diff options
Diffstat (limited to 'drivers/staging/media/hantro')
-rw-r--r-- | drivers/staging/media/hantro/Kconfig | 2 | ||||
-rw-r--r-- | drivers/staging/media/hantro/Makefile | 7 | ||||
-rw-r--r-- | drivers/staging/media/hantro/hantro.h | 51 | ||||
-rw-r--r-- | drivers/staging/media/hantro/hantro_drv.c | 106 | ||||
-rw-r--r-- | drivers/staging/media/hantro/hantro_g1_h264_dec.c | 292 | ||||
-rw-r--r-- | drivers/staging/media/hantro/hantro_g1_mpeg2_dec.c | 14 | ||||
-rw-r--r-- | drivers/staging/media/hantro/hantro_g1_vp8_dec.c | 503 | ||||
-rw-r--r-- | drivers/staging/media/hantro/hantro_h1_jpeg_enc.c | 11 | ||||
-rw-r--r-- | drivers/staging/media/hantro/hantro_h264.c | 646 | ||||
-rw-r--r-- | drivers/staging/media/hantro/hantro_hw.h | 80 | ||||
-rw-r--r-- | drivers/staging/media/hantro/hantro_v4l2.c | 11 | ||||
-rw-r--r-- | drivers/staging/media/hantro/hantro_vp8.c | 201 | ||||
-rw-r--r-- | drivers/staging/media/hantro/rk3288_vpu_hw.c | 41 | ||||
-rw-r--r-- | drivers/staging/media/hantro/rk3399_vpu_hw.c | 39 | ||||
-rw-r--r-- | drivers/staging/media/hantro/rk3399_vpu_hw_jpeg_enc.c | 12 | ||||
-rw-r--r-- | drivers/staging/media/hantro/rk3399_vpu_hw_mpeg2_dec.c | 14 | ||||
-rw-r--r-- | drivers/staging/media/hantro/rk3399_vpu_hw_vp8_dec.c | 595 |
17 files changed, 2565 insertions, 60 deletions
diff --git a/drivers/staging/media/hantro/Kconfig b/drivers/staging/media/hantro/Kconfig index be133bbaa68a..de77fe6554e7 100644 --- a/drivers/staging/media/hantro/Kconfig +++ b/drivers/staging/media/hantro/Kconfig @@ -20,4 +20,4 @@ config VIDEO_HANTRO_ROCKCHIP depends on ARCH_ROCKCHIP || COMPILE_TEST default y help - Enable support for RK3288 and RK3399 SoCs. + Enable support for RK3288, RK3328, and RK3399 SoCs. diff --git a/drivers/staging/media/hantro/Makefile b/drivers/staging/media/hantro/Makefile index 1584acdbf4a3..5d6b0383d280 100644 --- a/drivers/staging/media/hantro/Makefile +++ b/drivers/staging/media/hantro/Makefile @@ -4,11 +4,16 @@ hantro-vpu-y += \ hantro_drv.o \ hantro_v4l2.o \ hantro_h1_jpeg_enc.o \ + hantro_g1_h264_dec.o \ hantro_g1_mpeg2_dec.o \ + hantro_g1_vp8_dec.o \ rk3399_vpu_hw_jpeg_enc.o \ rk3399_vpu_hw_mpeg2_dec.o \ + rk3399_vpu_hw_vp8_dec.o \ hantro_jpeg.o \ - hantro_mpeg2.o + hantro_h264.o \ + hantro_mpeg2.o \ + hantro_vp8.o hantro-vpu-$(CONFIG_VIDEO_HANTRO_ROCKCHIP) += \ rk3288_vpu_hw.o \ diff --git a/drivers/staging/media/hantro/hantro.h b/drivers/staging/media/hantro/hantro.h index 62dcca9ff19c..f670bbde4159 100644 --- a/drivers/staging/media/hantro/hantro.h +++ b/drivers/staging/media/hantro/hantro.h @@ -20,11 +20,20 @@ #include <media/v4l2-ctrls.h> #include <media/v4l2-device.h> #include <media/v4l2-ioctl.h> +#include <media/v4l2-mem2mem.h> #include <media/videobuf2-core.h> #include <media/videobuf2-dma-contig.h> #include "hantro_hw.h" +#define VP8_MB_DIM 16 +#define VP8_MB_WIDTH(w) DIV_ROUND_UP(w, VP8_MB_DIM) +#define VP8_MB_HEIGHT(h) DIV_ROUND_UP(h, VP8_MB_DIM) + +#define H264_MB_DIM 16 +#define H264_MB_WIDTH(w) DIV_ROUND_UP(w, H264_MB_DIM) +#define H264_MB_HEIGHT(h) DIV_ROUND_UP(h, H264_MB_DIM) + #define MPEG2_MB_DIM 16 #define MPEG2_MB_WIDTH(w) DIV_ROUND_UP(w, MPEG2_MB_DIM) #define MPEG2_MB_HEIGHT(h) DIV_ROUND_UP(h, MPEG2_MB_DIM) @@ -38,8 +47,9 @@ struct hantro_codec_ops; #define HANTRO_JPEG_ENCODER BIT(0) #define HANTRO_ENCODERS 0x0000ffff - #define HANTRO_MPEG2_DECODER BIT(16) +#define HANTRO_VP8_DECODER BIT(17) +#define HANTRO_H264_DECODER BIT(18) #define HANTRO_DECODERS 0xffff0000 /** @@ -96,22 +106,24 @@ struct hantro_variant { * enum hantro_codec_mode - codec operating mode. * @HANTRO_MODE_NONE: No operating mode. Used for RAW video formats. * @HANTRO_MODE_JPEG_ENC: JPEG encoder. + * @HANTRO_MODE_H264_DEC: H264 decoder. * @HANTRO_MODE_MPEG2_DEC: MPEG-2 decoder. + * @HANTRO_MODE_VP8_DEC: VP8 decoder. */ enum hantro_codec_mode { HANTRO_MODE_NONE = -1, HANTRO_MODE_JPEG_ENC, + HANTRO_MODE_H264_DEC, HANTRO_MODE_MPEG2_DEC, + HANTRO_MODE_VP8_DEC, }; /* * struct hantro_ctrl - helper type to declare supported controls - * @id: V4L2 control ID (V4L2_CID_xxx) * @codec: codec id this control belong to (HANTRO_JPEG_ENCODER, etc.) * @cfg: control configuration */ struct hantro_ctrl { - unsigned int id; unsigned int codec; struct v4l2_ctrl_config cfg; }; @@ -215,6 +227,7 @@ struct hantro_dev { * @codec_ops: Set of operations related to codec mode. * @jpeg_enc: JPEG-encoding context. * @mpeg2_dec: MPEG-2-decoding context. + * @vp8_dec: VP8-decoding context. */ struct hantro_ctx { struct hantro_dev *dev; @@ -239,8 +252,10 @@ struct hantro_ctx { /* Specific for particular codec modes. */ union { + struct hantro_h264_dec_hw_ctx h264_dec; struct hantro_jpeg_enc_hw_ctx jpeg_enc; struct hantro_mpeg2_dec_hw_ctx mpeg2_dec; + struct hantro_vp8_dec_hw_ctx vp8_dec; }; }; @@ -265,6 +280,12 @@ struct hantro_fmt { struct v4l2_frmsize_stepwise frmsize; }; +struct hantro_reg { + u32 base; + u32 shift; + u32 mask; +}; + /* Logging helpers */ /** @@ -343,9 +364,33 @@ static inline u32 vdpu_read(struct hantro_dev *vpu, u32 reg) return val; } +static inline void hantro_reg_write(struct hantro_dev *vpu, + const struct hantro_reg *reg, + u32 val) +{ + u32 v; + + v = vdpu_read(vpu, reg->base); + v &= ~(reg->mask << reg->shift); + v |= ((val & reg->mask) << reg->shift); + vdpu_write_relaxed(vpu, v, reg->base); +} + bool hantro_is_encoder_ctx(const struct hantro_ctx *ctx); void *hantro_get_ctrl(struct hantro_ctx *ctx, u32 id); dma_addr_t hantro_get_ref(struct vb2_queue *q, u64 ts); +static inline struct vb2_v4l2_buffer * +hantro_get_src_buf(struct hantro_ctx *ctx) +{ + return v4l2_m2m_next_src_buf(ctx->fh.m2m_ctx); +} + +static inline struct vb2_v4l2_buffer * +hantro_get_dst_buf(struct hantro_ctx *ctx) +{ + return v4l2_m2m_next_dst_buf(ctx->fh.m2m_ctx); +} + #endif /* HANTRO_H_ */ diff --git a/drivers/staging/media/hantro/hantro_drv.c b/drivers/staging/media/hantro/hantro_drv.c index c3665f0e87a2..d8b6816b643b 100644 --- a/drivers/staging/media/hantro/hantro_drv.c +++ b/drivers/staging/media/hantro/hantro_drv.c @@ -111,8 +111,6 @@ static void hantro_job_finish(struct hantro_dev *vpu, src->sequence = ctx->sequence_out++; dst->sequence = ctx->sequence_cap++; - v4l2_m2m_buf_copy_metadata(src, dst, true); - ret = ctx->buf_finish(ctx, &dst->vb2_buf, bytesused); if (ret) result = VB2_BUF_STATE_ERROR; @@ -153,11 +151,37 @@ void hantro_watchdog(struct work_struct *work) } } +void hantro_prepare_run(struct hantro_ctx *ctx) +{ + struct vb2_v4l2_buffer *src_buf; + + src_buf = hantro_get_src_buf(ctx); + v4l2_ctrl_request_setup(src_buf->vb2_buf.req_obj.req, + &ctx->ctrl_handler); +} + +void hantro_finish_run(struct hantro_ctx *ctx) +{ + struct vb2_v4l2_buffer *src_buf; + + src_buf = hantro_get_src_buf(ctx); + v4l2_ctrl_request_complete(src_buf->vb2_buf.req_obj.req, + &ctx->ctrl_handler); + + /* Kick the watchdog. */ + schedule_delayed_work(&ctx->dev->watchdog_work, + msecs_to_jiffies(2000)); +} + static void device_run(void *priv) { struct hantro_ctx *ctx = priv; + struct vb2_v4l2_buffer *src, *dst; int ret; + src = hantro_get_src_buf(ctx); + dst = hantro_get_dst_buf(ctx); + ret = clk_bulk_enable(ctx->dev->variant->num_clocks, ctx->dev->clocks); if (ret) goto err_cancel_job; @@ -165,6 +189,8 @@ static void device_run(void *priv) if (ret < 0) goto err_cancel_job; + v4l2_m2m_buf_copy_metadata(src, dst, true); + ctx->codec_ops->run(ctx); return; @@ -262,28 +288,74 @@ static const struct v4l2_ctrl_ops hantro_ctrl_ops = { .s_ctrl = hantro_s_ctrl, }; -static struct hantro_ctrl controls[] = { +static const struct hantro_ctrl controls[] = { { - .id = V4L2_CID_JPEG_COMPRESSION_QUALITY, .codec = HANTRO_JPEG_ENCODER, .cfg = { + .id = V4L2_CID_JPEG_COMPRESSION_QUALITY, .min = 5, .max = 100, .step = 1, .def = 50, + .ops = &hantro_ctrl_ops, }, }, { - .id = V4L2_CID_MPEG_VIDEO_MPEG2_SLICE_PARAMS, .codec = HANTRO_MPEG2_DECODER, .cfg = { - .elem_size = sizeof(struct v4l2_ctrl_mpeg2_slice_params), + .id = V4L2_CID_MPEG_VIDEO_MPEG2_SLICE_PARAMS, }, }, { - .id = V4L2_CID_MPEG_VIDEO_MPEG2_QUANTIZATION, .codec = HANTRO_MPEG2_DECODER, .cfg = { - .elem_size = sizeof(struct v4l2_ctrl_mpeg2_quantization), + .id = V4L2_CID_MPEG_VIDEO_MPEG2_QUANTIZATION, + }, + }, { + .codec = HANTRO_VP8_DECODER, + .cfg = { + .id = V4L2_CID_MPEG_VIDEO_VP8_FRAME_HEADER, + }, + }, { + .codec = HANTRO_H264_DECODER, + .cfg = { + .id = V4L2_CID_MPEG_VIDEO_H264_DECODE_PARAMS, }, + }, { + .codec = HANTRO_H264_DECODER, + .cfg = { + .id = V4L2_CID_MPEG_VIDEO_H264_SLICE_PARAMS, + }, + }, { + .codec = HANTRO_H264_DECODER, + .cfg = { + .id = V4L2_CID_MPEG_VIDEO_H264_SPS, + }, + }, { + .codec = HANTRO_H264_DECODER, + .cfg = { + .id = V4L2_CID_MPEG_VIDEO_H264_PPS, + }, + }, { + .codec = HANTRO_H264_DECODER, + .cfg = { + .id = V4L2_CID_MPEG_VIDEO_H264_SCALING_MATRIX, + }, + }, { + .codec = HANTRO_H264_DECODER, + .cfg = { + .id = V4L2_CID_MPEG_VIDEO_H264_DECODE_MODE, + .min = V4L2_MPEG_VIDEO_H264_DECODE_MODE_FRAME_BASED, + .def = V4L2_MPEG_VIDEO_H264_DECODE_MODE_FRAME_BASED, + .max = V4L2_MPEG_VIDEO_H264_DECODE_MODE_FRAME_BASED, + }, + }, { + .codec = HANTRO_H264_DECODER, + .cfg = { + .id = V4L2_CID_MPEG_VIDEO_H264_START_CODE, + .min = V4L2_MPEG_VIDEO_H264_START_CODE_ANNEX_B, + .def = V4L2_MPEG_VIDEO_H264_START_CODE_ANNEX_B, + .max = V4L2_MPEG_VIDEO_H264_START_CODE_ANNEX_B, + }, + }, { }, }; @@ -298,22 +370,12 @@ static int hantro_ctrls_setup(struct hantro_dev *vpu, for (i = 0; i < num_ctrls; i++) { if (!(allowed_codecs & controls[i].codec)) continue; - if (!controls[i].cfg.elem_size) { - v4l2_ctrl_new_std(&ctx->ctrl_handler, - &hantro_ctrl_ops, - controls[i].id, controls[i].cfg.min, - controls[i].cfg.max, - controls[i].cfg.step, - controls[i].cfg.def); - } else { - controls[i].cfg.id = controls[i].id; - v4l2_ctrl_new_custom(&ctx->ctrl_handler, - &controls[i].cfg, NULL); - } + v4l2_ctrl_new_custom(&ctx->ctrl_handler, + &controls[i].cfg, NULL); if (ctx->ctrl_handler.error) { vpu_err("Adding control (%d) failed %d\n", - controls[i].id, + controls[i].cfg.id, ctx->ctrl_handler.error); v4l2_ctrl_handler_free(&ctx->ctrl_handler); return ctx->ctrl_handler.error; @@ -419,6 +481,7 @@ static const struct v4l2_file_operations hantro_fops = { static const struct of_device_id of_hantro_match[] = { #ifdef CONFIG_VIDEO_HANTRO_ROCKCHIP { .compatible = "rockchip,rk3399-vpu", .data = &rk3399_vpu_variant, }, + { .compatible = "rockchip,rk3328-vpu", .data = &rk3328_vpu_variant, }, { .compatible = "rockchip,rk3288-vpu", .data = &rk3288_vpu_variant, }, #endif { /* sentinel */ } @@ -724,6 +787,7 @@ static int hantro_probe(struct platform_device *pdev) dev_err(vpu->dev, "Could not set DMA coherent mask.\n"); return ret; } + vb2_dma_contig_set_max_seg_size(&pdev->dev, DMA_BIT_MASK(32)); for (i = 0; i < vpu->variant->num_irqs; i++) { const char *irq_name = vpu->variant->irqs[i].name; diff --git a/drivers/staging/media/hantro/hantro_g1_h264_dec.c b/drivers/staging/media/hantro/hantro_g1_h264_dec.c new file mode 100644 index 000000000000..7ab534936843 --- /dev/null +++ b/drivers/staging/media/hantro/hantro_g1_h264_dec.c @@ -0,0 +1,292 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Rockchip RK3288 VPU codec driver + * + * Copyright (c) 2014 Rockchip Electronics Co., Ltd. + * Hertz Wong <hertz.wong@rock-chips.com> + * Herman Chen <herman.chen@rock-chips.com> + * + * Copyright (C) 2014 Google, Inc. + * Tomasz Figa <tfiga@chromium.org> + */ + +#include <linux/types.h> +#include <linux/sort.h> + +#include <media/v4l2-mem2mem.h> + +#include "hantro_g1_regs.h" +#include "hantro_hw.h" +#include "hantro_v4l2.h" + +static void set_params(struct hantro_ctx *ctx) +{ + const struct hantro_h264_dec_ctrls *ctrls = &ctx->h264_dec.ctrls; + const struct v4l2_ctrl_h264_decode_params *dec_param = ctrls->decode; + const struct v4l2_ctrl_h264_slice_params *slices = ctrls->slices; + const struct v4l2_ctrl_h264_sps *sps = ctrls->sps; + const struct v4l2_ctrl_h264_pps *pps = ctrls->pps; + struct vb2_v4l2_buffer *src_buf = hantro_get_src_buf(ctx); + struct hantro_dev *vpu = ctx->dev; + u32 reg; + + /* Decoder control register 0. */ + reg = G1_REG_DEC_CTRL0_DEC_AXI_WR_ID(0x0); + if (sps->flags & V4L2_H264_SPS_FLAG_MB_ADAPTIVE_FRAME_FIELD) + reg |= G1_REG_DEC_CTRL0_SEQ_MBAFF_E; + reg |= G1_REG_DEC_CTRL0_PICORD_COUNT_E; + if (dec_param->nal_ref_idc) + reg |= G1_REG_DEC_CTRL0_WRITE_MVS_E; + + if (!(sps->flags & V4L2_H264_SPS_FLAG_FRAME_MBS_ONLY) && + (sps->flags & V4L2_H264_SPS_FLAG_MB_ADAPTIVE_FRAME_FIELD || + slices[0].flags & V4L2_H264_SLICE_FLAG_FIELD_PIC)) + reg |= G1_REG_DEC_CTRL0_PIC_INTERLACE_E; + if (slices[0].flags & V4L2_H264_SLICE_FLAG_FIELD_PIC) + reg |= G1_REG_DEC_CTRL0_PIC_FIELDMODE_E; + if (!(slices[0].flags & V4L2_H264_SLICE_FLAG_BOTTOM_FIELD)) + reg |= G1_REG_DEC_CTRL0_PIC_TOPFIELD_E; + vdpu_write_relaxed(vpu, reg, G1_REG_DEC_CTRL0); + + /* Decoder control register 1. */ + reg = G1_REG_DEC_CTRL1_PIC_MB_WIDTH(sps->pic_width_in_mbs_minus1 + 1) | + G1_REG_DEC_CTRL1_PIC_MB_HEIGHT_P(sps->pic_height_in_map_units_minus1 + 1) | + G1_REG_DEC_CTRL1_REF_FRAMES(sps->max_num_ref_frames); + vdpu_write_relaxed(vpu, reg, G1_REG_DEC_CTRL1); + + /* Decoder control register 2. */ + reg = G1_REG_DEC_CTRL2_CH_QP_OFFSET(pps->chroma_qp_index_offset) | + G1_REG_DEC_CTRL2_CH_QP_OFFSET2(pps->second_chroma_qp_index_offset); + + /* always use the matrix sent from userspace */ + reg |= G1_REG_DEC_CTRL2_TYPE1_QUANT_E; + + if (slices[0].flags & V4L2_H264_SLICE_FLAG_FIELD_PIC) + reg |= G1_REG_DEC_CTRL2_FIELDPIC_FLAG_E; + vdpu_write_relaxed(vpu, reg, G1_REG_DEC_CTRL2); + + /* Decoder control register 3. */ + reg = G1_REG_DEC_CTRL3_START_CODE_E | + G1_REG_DEC_CTRL3_INIT_QP(pps->pic_init_qp_minus26 + 26) | + G1_REG_DEC_CTRL3_STREAM_LEN(vb2_get_plane_payload(&src_buf->vb2_buf, 0)); + vdpu_write_relaxed(vpu, reg, G1_REG_DEC_CTRL3); + + /* Decoder control register 4. */ + reg = G1_REG_DEC_CTRL4_FRAMENUM_LEN(sps->log2_max_frame_num_minus4 + 4) | + G1_REG_DEC_CTRL4_FRAMENUM(slices[0].frame_num) | + G1_REG_DEC_CTRL4_WEIGHT_BIPR_IDC(pps->weighted_bipred_idc); + if (pps->flags & V4L2_H264_PPS_FLAG_ENTROPY_CODING_MODE) + reg |= G1_REG_DEC_CTRL4_CABAC_E; + if (sps->flags & V4L2_H264_SPS_FLAG_DIRECT_8X8_INFERENCE) + reg |= G1_REG_DEC_CTRL4_DIR_8X8_INFER_E; + if (sps->chroma_format_idc == 0) + reg |= G1_REG_DEC_CTRL4_BLACKWHITE_E; + if (pps->flags & V4L2_H264_PPS_FLAG_WEIGHTED_PRED) + reg |= G1_REG_DEC_CTRL4_WEIGHT_PRED_E; + vdpu_write_relaxed(vpu, reg, G1_REG_DEC_CTRL4); + + /* Decoder control register 5. */ + reg = G1_REG_DEC_CTRL5_REFPIC_MK_LEN(slices[0].dec_ref_pic_marking_bit_size) | + G1_REG_DEC_CTRL5_IDR_PIC_ID(slices[0].idr_pic_id); + if (pps->flags & V4L2_H264_PPS_FLAG_CONSTRAINED_INTRA_PRED) + reg |= G1_REG_DEC_CTRL5_CONST_INTRA_E; + if (pps->flags & V4L2_H264_PPS_FLAG_DEBLOCKING_FILTER_CONTROL_PRESENT) + reg |= G1_REG_DEC_CTRL5_FILT_CTRL_PRES; + if (pps->flags & V4L2_H264_PPS_FLAG_REDUNDANT_PIC_CNT_PRESENT) + reg |= G1_REG_DEC_CTRL5_RDPIC_CNT_PRES; + if (pps->flags & V4L2_H264_PPS_FLAG_TRANSFORM_8X8_MODE) + reg |= G1_REG_DEC_CTRL5_8X8TRANS_FLAG_E; + if (dec_param->flags & V4L2_H264_DECODE_PARAM_FLAG_IDR_PIC) + reg |= G1_REG_DEC_CTRL5_IDR_PIC_E; + vdpu_write_relaxed(vpu, reg, G1_REG_DEC_CTRL5); + + /* Decoder control register 6. */ + reg = G1_REG_DEC_CTRL6_PPS_ID(slices[0].pic_parameter_set_id) | + G1_REG_DEC_CTRL6_REFIDX0_ACTIVE(pps->num_ref_idx_l0_default_active_minus1 + 1) | + G1_REG_DEC_CTRL6_REFIDX1_ACTIVE(pps->num_ref_idx_l1_default_active_minus1 + 1) | + G1_REG_DEC_CTRL6_POC_LENGTH(slices[0].pic_order_cnt_bit_size); + vdpu_write_relaxed(vpu, reg, G1_REG_DEC_CTRL6); + + /* Error concealment register. */ + vdpu_write_relaxed(vpu, 0, G1_REG_ERR_CONC); + + /* Prediction filter tap register. */ + vdpu_write_relaxed(vpu, + G1_REG_PRED_FLT_PRED_BC_TAP_0_0(1) | + G1_REG_PRED_FLT_PRED_BC_TAP_0_1(-5 & 0x3ff) | + G1_REG_PRED_FLT_PRED_BC_TAP_0_2(20), + G1_REG_PRED_FLT); + + /* Reference picture buffer control register. */ + vdpu_write_relaxed(vpu, 0, G1_REG_REF_BUF_CTRL); + + /* Reference picture buffer control register 2. */ + vdpu_write_relaxed(vpu, G1_REG_REF_BUF_CTRL2_APF_THRESHOLD(8), + G1_REG_REF_BUF_CTRL2); +} + +static void set_ref(struct hantro_ctx *ctx) +{ + struct v4l2_h264_dpb_entry *dpb = ctx->h264_dec.dpb; + const u8 *b0_reflist, *b1_reflist, *p_reflist; + struct hantro_dev *vpu = ctx->dev; + u32 dpb_longterm = 0; + u32 dpb_valid = 0; + int reg_num; + u32 reg; + int i; + + /* + * Set up bit maps of valid and long term DPBs. + * NOTE: The bits are reversed, i.e. MSb is DPB 0. + */ + for (i = 0; i < HANTRO_H264_DPB_SIZE; ++i) { + if (dpb[i].flags & V4L2_H264_DPB_ENTRY_FLAG_ACTIVE) + dpb_valid |= BIT(HANTRO_H264_DPB_SIZE - 1 - i); + + if (dpb[i].flags & V4L2_H264_DPB_ENTRY_FLAG_LONG_TERM) + dpb_longterm |= BIT(HANTRO_H264_DPB_SIZE - 1 - i); + } + vdpu_write_relaxed(vpu, dpb_valid << 16, G1_REG_VALID_REF); + vdpu_write_relaxed(vpu, dpb_longterm << 16, G1_REG_LT_REF); + + /* + * Set up reference frame picture numbers. + * + * Each G1_REG_REF_PIC(x) register contains numbers of two + * subsequential reference pictures. + */ + for (i = 0; i < HANTRO_H264_DPB_SIZE; i += 2) { + reg = 0; + if (dpb[i].flags & V4L2_H264_DPB_ENTRY_FLAG_LONG_TERM) + reg |= G1_REG_REF_PIC_REFER0_NBR(dpb[i].pic_num); + else + reg |= G1_REG_REF_PIC_REFER0_NBR(dpb[i].frame_num); + + if (dpb[i + 1].flags & V4L2_H264_DPB_ENTRY_FLAG_LONG_TERM) + reg |= G1_REG_REF_PIC_REFER1_NBR(dpb[i + 1].pic_num); + else + reg |= G1_REG_REF_PIC_REFER1_NBR(dpb[i + 1].frame_num); + + vdpu_write_relaxed(vpu, reg, G1_REG_REF_PIC(i / 2)); + } + + b0_reflist = ctx->h264_dec.reflists.b0; + b1_reflist = ctx->h264_dec.reflists.b1; + p_reflist = ctx->h264_dec.reflists.p; + + /* + * Each G1_REG_BD_REF_PIC(x) register contains three entries + * of each forward and backward picture list. + */ + reg_num = 0; + for (i = 0; i < 15; i += 3) { + reg = G1_REG_BD_REF_PIC_BINIT_RLIST_F0(b0_reflist[i]) | + G1_REG_BD_REF_PIC_BINIT_RLIST_F1(b0_reflist[i + 1]) | + G1_REG_BD_REF_PIC_BINIT_RLIST_F2(b0_reflist[i + 2]) | + G1_REG_BD_REF_PIC_BINIT_RLIST_B0(b1_reflist[i]) | + G1_REG_BD_REF_PIC_BINIT_RLIST_B1(b1_reflist[i + 1]) | + G1_REG_BD_REF_PIC_BINIT_RLIST_B2(b1_reflist[i + 2]); + vdpu_write_relaxed(vpu, reg, G1_REG_BD_REF_PIC(reg_num++)); + } + + /* + * G1_REG_BD_P_REF_PIC register contains last entries (index 15) + * of forward and backward reference picture lists and first 4 entries + * of P forward picture list. + */ + reg = G1_REG_BD_P_REF_PIC_BINIT_RLIST_F15(b0_reflist[15]) | + G1_REG_BD_P_REF_PIC_BINIT_RLIST_B15(b1_reflist[15]) | + G1_REG_BD_P_REF_PIC_PINIT_RLIST_F0(p_reflist[0]) | + G1_REG_BD_P_REF_PIC_PINIT_RLIST_F1(p_reflist[1]) | + G1_REG_BD_P_REF_PIC_PINIT_RLIST_F2(p_reflist[2]) | + G1_REG_BD_P_REF_PIC_PINIT_RLIST_F3(p_reflist[3]); + vdpu_write_relaxed(vpu, reg, G1_REG_BD_P_REF_PIC); + + /* + * Each G1_REG_FWD_PIC(x) register contains six consecutive + * entries of P forward picture list, starting from index 4. + */ + reg_num = 0; + for (i = 4; i < HANTRO_H264_DPB_SIZE; i += 6) { + reg = G1_REG_FWD_PIC_PINIT_RLIST_F0(p_reflist[i]) | + G1_REG_FWD_PIC_PINIT_RLIST_F1(p_reflist[i + 1]) | + G1_REG_FWD_PIC_PINIT_RLIST_F2(p_reflist[i + 2]) | + G1_REG_FWD_PIC_PINIT_RLIST_F3(p_reflist[i + 3]) | + G1_REG_FWD_PIC_PINIT_RLIST_F4(p_reflist[i + 4]) | + G1_REG_FWD_PIC_PINIT_RLIST_F5(p_reflist[i + 5]); + vdpu_write_relaxed(vpu, reg, G1_REG_FWD_PIC(reg_num++)); + } + + /* Set up addresses of DPB buffers. */ + for (i = 0; i < HANTRO_H264_DPB_SIZE; i++) { + struct vb2_buffer *buf = hantro_h264_get_ref_buf(ctx, i); + + vdpu_write_relaxed(vpu, vb2_dma_contig_plane_dma_addr(buf, 0), + G1_REG_ADDR_REF(i)); + } +} + +static void set_buffers(struct hantro_ctx *ctx) +{ + const struct hantro_h264_dec_ctrls *ctrls = &ctx->h264_dec.ctrls; + struct vb2_v4l2_buffer *src_buf, *dst_buf; + struct hantro_dev *vpu = ctx->dev; + dma_addr_t src_dma, dst_dma; + + src_buf = hantro_get_src_buf(ctx); + dst_buf = hantro_get_dst_buf(ctx); + + /* Source (stream) buffer. */ + src_dma = vb2_dma_contig_plane_dma_addr(&src_buf->vb2_buf, 0); + vdpu_write_relaxed(vpu, src_dma, G1_REG_ADDR_STR); + + /* Destination (decoded frame) buffer. */ + dst_dma = vb2_dma_contig_plane_dma_addr(&dst_buf->vb2_buf, 0); + vdpu_write_relaxed(vpu, dst_dma, G1_REG_ADDR_DST); + + /* Higher profiles require DMV buffer appended to reference frames. */ + if (ctrls->sps->profile_idc > 66) { + size_t pic_size = ctx->h264_dec.pic_size; + size_t mv_offset = round_up(pic_size, 8); + + if (ctrls->slices[0].flags & V4L2_H264_SLICE_FLAG_BOTTOM_FIELD) + mv_offset += 32 * H264_MB_WIDTH(ctx->dst_fmt.width); + + vdpu_write_relaxed(vpu, dst_dma + mv_offset, + G1_REG_ADDR_DIR_MV); + } + + /* Auxiliary buffer prepared in hantro_g1_h264_dec_prepare_table(). */ + vdpu_write_relaxed(vpu, ctx->h264_dec.priv.dma, G1_REG_ADDR_QTABLE); +} + +void hantro_g1_h264_dec_run(struct hantro_ctx *ctx) +{ + struct hantro_dev *vpu = ctx->dev; + + /* Prepare the H264 decoder context. */ + if (hantro_h264_dec_prepare_run(ctx)) + return; + + /* Configure hardware registers. */ + set_params(ctx); + set_ref(ctx); + set_buffers(ctx); + + hantro_finish_run(ctx); + + /* Start decoding! */ + vdpu_write_relaxed(vpu, + G1_REG_CONFIG_DEC_AXI_RD_ID(0xffu) | + G1_REG_CONFIG_DEC_TIMEOUT_E | + G1_REG_CONFIG_DEC_OUT_ENDIAN | + G1_REG_CONFIG_DEC_STRENDIAN_E | + G1_REG_CONFIG_DEC_MAX_BURST(16) | + G1_REG_CONFIG_DEC_OUTSWAP32_E | + G1_REG_CONFIG_DEC_INSWAP32_E | + G1_REG_CONFIG_DEC_STRSWAP32_E | + G1_REG_CONFIG_DEC_CLK_GATE_E, + G1_REG_CONFIG); + vdpu_write(vpu, G1_REG_INTERRUPT_DEC_E, G1_REG_INTERRUPT); +} diff --git a/drivers/staging/media/hantro/hantro_g1_mpeg2_dec.c b/drivers/staging/media/hantro/hantro_g1_mpeg2_dec.c index e592c1b66375..80f0e94f8afa 100644 --- a/drivers/staging/media/hantro/hantro_g1_mpeg2_dec.c +++ b/drivers/staging/media/hantro/hantro_g1_mpeg2_dec.c @@ -167,12 +167,11 @@ void hantro_g1_mpeg2_dec_run(struct hantro_ctx *ctx) const struct v4l2_mpeg2_picture *picture; u32 reg; - src_buf = v4l2_m2m_next_src_buf(ctx->fh.m2m_ctx); - dst_buf = v4l2_m2m_next_dst_buf(ctx->fh.m2m_ctx); + src_buf = hantro_get_src_buf(ctx); + dst_buf = hantro_get_dst_buf(ctx); /* Apply request controls if any */ - v4l2_ctrl_request_setup(src_buf->vb2_buf.req_obj.req, - &ctx->ctrl_handler); + hantro_prepare_run(ctx); slice_params = hantro_get_ctrl(ctx, V4L2_CID_MPEG_VIDEO_MPEG2_SLICE_PARAMS); @@ -248,12 +247,7 @@ void hantro_g1_mpeg2_dec_run(struct hantro_ctx *ctx) &dst_buf->vb2_buf, sequence, picture, slice_params); - /* Controls no longer in-use, we can complete them */ - v4l2_ctrl_request_complete(src_buf->vb2_buf.req_obj.req, - &ctx->ctrl_handler); - - /* Kick the watchdog and start decoding */ - schedule_delayed_work(&vpu->watchdog_work, msecs_to_jiffies(2000)); + hantro_finish_run(ctx); reg = G1_REG_DEC_E(1); vdpu_write(vpu, reg, G1_SWREG(1)); diff --git a/drivers/staging/media/hantro/hantro_g1_vp8_dec.c b/drivers/staging/media/hantro/hantro_g1_vp8_dec.c new file mode 100644 index 000000000000..6d99c2be01cf --- /dev/null +++ b/drivers/staging/media/hantro/hantro_g1_vp8_dec.c @@ -0,0 +1,503 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Hantro VP8 codec driver + * + * Copyright (C) 2019 Rockchip Electronics Co., Ltd. + * ZhiChao Yu <zhichao.yu@rock-chips.com> + * + * Copyright (C) 2019 Google, Inc. + * Tomasz Figa <tfiga@chromium.org> + */ + +#include <media/v4l2-mem2mem.h> +#include <media/vp8-ctrls.h> + +#include "hantro_hw.h" +#include "hantro.h" +#include "hantro_g1_regs.h" + +/* DCT partition base address regs */ +static const struct hantro_reg vp8_dec_dct_base[8] = { + { G1_REG_ADDR_STR, 0, 0xffffffff }, + { G1_REG_ADDR_REF(8), 0, 0xffffffff }, + { G1_REG_ADDR_REF(9), 0, 0xffffffff }, + { G1_REG_ADDR_REF(10), 0, 0xffffffff }, + { G1_REG_ADDR_REF(11), 0, 0xffffffff }, + { G1_REG_ADDR_REF(12), 0, 0xffffffff }, + { G1_REG_ADDR_REF(14), 0, 0xffffffff }, + { G1_REG_ADDR_REF(15), 0, 0xffffffff }, +}; + +/* Loop filter level regs */ +static const struct hantro_reg vp8_dec_lf_level[4] = { + { G1_REG_REF_PIC(2), 18, 0x3f }, + { G1_REG_REF_PIC(2), 12, 0x3f }, + { G1_REG_REF_PIC(2), 6, 0x3f }, + { G1_REG_REF_PIC(2), 0, 0x3f }, +}; + +/* Macroblock loop filter level adjustment regs */ +static const struct hantro_reg vp8_dec_mb_adj[4] = { + { G1_REG_REF_PIC(0), 21, 0x7f }, + { G1_REG_REF_PIC(0), 14, 0x7f }, + { G1_REG_REF_PIC(0), 7, 0x7f }, + { G1_REG_REF_PIC(0), 0, 0x7f }, +}; + +/* Reference frame adjustment regs */ +static const struct hantro_reg vp8_dec_ref_adj[4] = { + { G1_REG_REF_PIC(1), 21, 0x7f }, + { G1_REG_REF_PIC(1), 14, 0x7f }, + { G1_REG_REF_PIC(1), 7, 0x7f }, + { G1_REG_REF_PIC(1), 0, 0x7f }, +}; + +/* Quantizer */ +static const struct hantro_reg vp8_dec_quant[4] = { + { G1_REG_REF_PIC(3), 11, 0x7ff }, + { G1_REG_REF_PIC(3), 0, 0x7ff }, + { G1_REG_BD_REF_PIC(4), 11, 0x7ff }, + { G1_REG_BD_REF_PIC(4), 0, 0x7ff }, +}; + +/* Quantizer delta regs */ +static const struct hantro_reg vp8_dec_quant_delta[5] = { + { G1_REG_REF_PIC(3), 27, 0x1f }, + { G1_REG_REF_PIC(3), 22, 0x1f }, + { G1_REG_BD_REF_PIC(4), 27, 0x1f }, + { G1_REG_BD_REF_PIC(4), 22, 0x1f }, + { G1_REG_BD_P_REF_PIC, 27, 0x1f }, +}; + +/* DCT partition start bits regs */ +static const struct hantro_reg vp8_dec_dct_start_bits[8] = { + { G1_REG_DEC_CTRL2, 26, 0x3f }, { G1_REG_DEC_CTRL4, 26, 0x3f }, + { G1_REG_DEC_CTRL4, 20, 0x3f }, { G1_REG_DEC_CTRL7, 24, 0x3f }, + { G1_REG_DEC_CTRL7, 18, 0x3f }, { G1_REG_DEC_CTRL7, 12, 0x3f }, + { G1_REG_DEC_CTRL7, 6, 0x3f }, { G1_REG_DEC_CTRL7, 0, 0x3f }, +}; + +/* Precision filter tap regs */ +static const struct hantro_reg vp8_dec_pred_bc_tap[8][4] = { + { + { G1_REG_PRED_FLT, 22, 0x3ff }, + { G1_REG_PRED_FLT, 12, 0x3ff }, + { G1_REG_PRED_FLT, 2, 0x3ff }, + { G1_REG_REF_PIC(4), 22, 0x3ff }, + }, + { + { G1_REG_REF_PIC(4), 12, 0x3ff }, + { G1_REG_REF_PIC(4), 2, 0x3ff }, + { G1_REG_REF_PIC(5), 22, 0x3ff }, + { G1_REG_REF_PIC(5), 12, 0x3ff }, + }, + { + { G1_REG_REF_PIC(5), 2, 0x3ff }, + { G1_REG_REF_PIC(6), 22, 0x3ff }, + { G1_REG_REF_PIC(6), 12, 0x3ff }, + { G1_REG_REF_PIC(6), 2, 0x3ff }, + }, + { + { G1_REG_REF_PIC(7), 22, 0x3ff }, + { G1_REG_REF_PIC(7), 12, 0x3ff }, + { G1_REG_REF_PIC(7), 2, 0x3ff }, + { G1_REG_LT_REF, 22, 0x3ff }, + }, + { + { G1_REG_LT_REF, 12, 0x3ff }, + { G1_REG_LT_REF, 2, 0x3ff }, + { G1_REG_VALID_REF, 22, 0x3ff }, + { G1_REG_VALID_REF, 12, 0x3ff }, + }, + { + { G1_REG_VALID_REF, 2, 0x3ff }, + { G1_REG_BD_REF_PIC(0), 22, 0x3ff }, + { G1_REG_BD_REF_PIC(0), 12, 0x3ff }, + { G1_REG_BD_REF_PIC(0), 2, 0x3ff }, + }, + { + { G1_REG_BD_REF_PIC(1), 22, 0x3ff }, + { G1_REG_BD_REF_PIC(1), 12, 0x3ff }, + { G1_REG_BD_REF_PIC(1), 2, 0x3ff }, + { G1_REG_BD_REF_PIC(2), 22, 0x3ff }, + }, + { + { G1_REG_BD_REF_PIC(2), 12, 0x3ff }, + { G1_REG_BD_REF_PIC(2), 2, 0x3ff }, + { G1_REG_BD_REF_PIC(3), 22, 0x3ff }, + { G1_REG_BD_REF_PIC(3), 12, 0x3ff }, + }, +}; + +/* + * Set loop filters + */ +static void cfg_lf(struct hantro_ctx *ctx, + const struct v4l2_ctrl_vp8_frame_header *hdr) +{ + const struct v4l2_vp8_segment_header *seg = &hdr->segment_header; + const struct v4l2_vp8_loopfilter_header *lf = &hdr->lf_header; + struct hantro_dev *vpu = ctx->dev; + unsigned int i; + u32 reg; + + if (!(seg->flags & V4L2_VP8_SEGMENT_HEADER_FLAG_ENABLED)) { + hantro_reg_write(vpu, &vp8_dec_lf_level[0], lf->level); + } else if (seg->flags & V4L2_VP8_SEGMENT_HEADER_FLAG_DELTA_VALUE_MODE) { + for (i = 0; i < 4; i++) { + u32 lf_level = clamp(lf->level + seg->lf_update[i], + 0, 63); + + hantro_reg_write(vpu, &vp8_dec_lf_level[i], lf_level); + } + } else { + for (i = 0; i < 4; i++) + hantro_reg_write(vpu, &vp8_dec_lf_level[i], + seg->lf_update[i]); + } + + reg = G1_REG_REF_PIC_FILT_SHARPNESS(lf->sharpness_level); + if (lf->flags & V4L2_VP8_LF_FILTER_TYPE_SIMPLE) + reg |= G1_REG_REF_PIC_FILT_TYPE_E; + vdpu_write_relaxed(vpu, reg, G1_REG_REF_PIC(0)); + + if (lf->flags & V4L2_VP8_LF_HEADER_ADJ_ENABLE) { + for (i = 0; i < 4; i++) { + hantro_reg_write(vpu, &vp8_dec_mb_adj[i], + lf->mb_mode_delta[i]); + hantro_reg_write(vpu, &vp8_dec_ref_adj[i], + lf->ref_frm_delta[i]); + } + } +} + +/* + * Set quantization parameters + */ +static void cfg_qp(struct hantro_ctx *ctx, + const struct v4l2_ctrl_vp8_frame_header *hdr) +{ + const struct v4l2_vp8_quantization_header *q = &hdr->quant_header; + const struct v4l2_vp8_segment_header *seg = &hdr->segment_header; + struct hantro_dev *vpu = ctx->dev; + unsigned int i; + + if (!(seg->flags & V4L2_VP8_SEGMENT_HEADER_FLAG_ENABLED)) { + hantro_reg_write(vpu, &vp8_dec_quant[0], q->y_ac_qi); + } else if (seg->flags & V4L2_VP8_SEGMENT_HEADER_FLAG_DELTA_VALUE_MODE) { + for (i = 0; i < 4; i++) { + u32 quant = clamp(q->y_ac_qi + seg->quant_update[i], + 0, 127); + + hantro_reg_write(vpu, &vp8_dec_quant[i], quant); + } + } else { + for (i = 0; i < 4; i++) + hantro_reg_write(vpu, &vp8_dec_quant[i], + seg->quant_update[i]); + } + + hantro_reg_write(vpu, &vp8_dec_quant_delta[0], q->y_dc_delta); + hantro_reg_write(vpu, &vp8_dec_quant_delta[1], q->y2_dc_delta); + hantro_reg_write(vpu, &vp8_dec_quant_delta[2], q->y2_ac_delta); + hantro_reg_write(vpu, &vp8_dec_quant_delta[3], q->uv_dc_delta); + hantro_reg_write(vpu, &vp8_dec_quant_delta[4], q->uv_ac_delta); +} + +/* + * set control partition and DCT partition regs + * + * VP8 frame stream data layout: + * + * first_part_size parttion_sizes[0] + * ^ ^ + * src_dma | | + * ^ +--------+------+ +-----+-----+ + * | | control part | | | + * +--------+----------------+------------------+-----------+-----+-----------+ + * | tag 3B | extra 7B | hdr | mb_data | DCT sz | DCT part0 | ... | DCT partn | + * +--------+-----------------------------------+-----------+-----+-----------+ + * | | | | + * v +----+---+ v + * mb_start | src_dma_end + * v + * DCT size part + * (num_dct-1)*3B + * Note: + * 1. only key-frames have extra 7-bytes + * 2. all offsets are base on src_dma + * 3. number of DCT parts is 1, 2, 4 or 8 + * 4. the addresses set to the VPU must be 64-bits aligned + */ +static void cfg_parts(struct hantro_ctx *ctx, + const struct v4l2_ctrl_vp8_frame_header *hdr) +{ + struct hantro_dev *vpu = ctx->dev; + struct vb2_v4l2_buffer *vb2_src; + u32 first_part_offset = VP8_FRAME_IS_KEY_FRAME(hdr) ? 10 : 3; + u32 mb_size, mb_offset_bytes, mb_offset_bits, mb_start_bits; + u32 dct_size_part_size, dct_part_offset; + struct hantro_reg reg; + dma_addr_t src_dma; + u32 dct_part_total_len = 0; + u32 count = 0; + unsigned int i; + + vb2_src = hantro_get_src_buf(ctx); + src_dma = vb2_dma_contig_plane_dma_addr(&vb2_src->vb2_buf, 0); + + /* + * Calculate control partition mb data info + * @first_part_header_bits: bits offset of mb data from first + * part start pos + * @mb_offset_bits: bits offset of mb data from src_dma + * base addr + * @mb_offset_byte: bytes offset of mb data from src_dma + * base addr + * @mb_start_bits: bits offset of mb data from mb data + * 64bits alignment addr + */ + mb_offset_bits = first_part_offset * 8 + + hdr->first_part_header_bits + 8; + mb_offset_bytes = mb_offset_bits / 8; + mb_start_bits = mb_offset_bits - + (mb_offset_bytes & (~DEC_8190_ALIGN_MASK)) * 8; + mb_size = hdr->first_part_size - + (mb_offset_bytes - first_part_offset) + + (mb_offset_bytes & DEC_8190_ALIGN_MASK); + + /* Macroblock data aligned base addr */ + vdpu_write_relaxed(vpu, (mb_offset_bytes & (~DEC_8190_ALIGN_MASK)) + + src_dma, G1_REG_ADDR_REF(13)); + + /* Macroblock data start bits */ + reg.base = G1_REG_DEC_CTRL2; + reg.mask = 0x3f; + reg.shift = 18; + hantro_reg_write(vpu, ®, mb_start_bits); + + /* Macroblock aligned data length */ + reg.base = G1_REG_DEC_CTRL6; + reg.mask = 0x3fffff; + reg.shift = 0; + hantro_reg_write(vpu, ®, mb_size + 1); + + /* + * Calculate DCT partition info + * @dct_size_part_size: Containing sizes of DCT part, every DCT part + * has 3 bytes to store its size, except the last + * DCT part + * @dct_part_offset: bytes offset of DCT parts from src_dma base addr + * @dct_part_total_len: total size of all DCT parts + */ + dct_size_part_size = (hdr->num_dct_parts - 1) * 3; + dct_part_offset = first_part_offset + hdr->first_part_size; + for (i = 0; i < hdr->num_dct_parts; i++) + dct_part_total_len += hdr->dct_part_sizes[i]; + dct_part_total_len += dct_size_part_size; + dct_part_total_len += (dct_part_offset & DEC_8190_ALIGN_MASK); + + /* Number of DCT partitions */ + reg.base = G1_REG_DEC_CTRL6; + reg.mask = 0xf; + reg.shift = 24; + hantro_reg_write(vpu, ®, hdr->num_dct_parts - 1); + + /* DCT partition length */ + vdpu_write_relaxed(vpu, + G1_REG_DEC_CTRL3_STREAM_LEN(dct_part_total_len), + G1_REG_DEC_CTRL3); + + /* DCT partitions base address */ + for (i = 0; i < hdr->num_dct_parts; i++) { + u32 byte_offset = dct_part_offset + dct_size_part_size + count; + u32 base_addr = byte_offset + src_dma; + + hantro_reg_write(vpu, &vp8_dec_dct_base[i], + base_addr & (~DEC_8190_ALIGN_MASK)); + + hantro_reg_write(vpu, &vp8_dec_dct_start_bits[i], + (byte_offset & DEC_8190_ALIGN_MASK) * 8); + + count += hdr->dct_part_sizes[i]; + } +} + +/* + * prediction filter taps + * normal 6-tap filters + */ +static void cfg_tap(struct hantro_ctx *ctx, + const struct v4l2_ctrl_vp8_frame_header *hdr) +{ + struct hantro_dev *vpu = ctx->dev; + struct hantro_reg reg; + u32 val = 0; + int i, j; + + reg.base = G1_REG_BD_REF_PIC(3); + reg.mask = 0xf; + + if ((hdr->version & 0x03) != 0) + return; /* Tap filter not used. */ + + for (i = 0; i < 8; i++) { + val = (hantro_vp8_dec_mc_filter[i][0] << 2) | + hantro_vp8_dec_mc_filter[i][5]; + + for (j = 0; j < 4; j++) + hantro_reg_write(vpu, &vp8_dec_pred_bc_tap[i][j], + hantro_vp8_dec_mc_filter[i][j + 1]); + + switch (i) { + case 2: + reg.shift = 8; + break; + case 4: + reg.shift = 4; + break; + case 6: + reg.shift = 0; + break; + default: + continue; + } + + hantro_reg_write(vpu, ®, val); + } +} + +static void cfg_ref(struct hantro_ctx *ctx, + const struct v4l2_ctrl_vp8_frame_header *hdr) +{ + struct vb2_queue *cap_q = &ctx->fh.m2m_ctx->cap_q_ctx.q; + struct hantro_dev *vpu = ctx->dev; + struct vb2_v4l2_buffer *vb2_dst; + dma_addr_t ref; + + vb2_dst = hantro_get_dst_buf(ctx); + + ref = hantro_get_ref(cap_q, hdr->last_frame_ts); + if (!ref) + ref = vb2_dma_contig_plane_dma_addr(&vb2_dst->vb2_buf, 0); + vdpu_write_relaxed(vpu, ref, G1_REG_ADDR_REF(0)); + + ref = hantro_get_ref(cap_q, hdr->golden_frame_ts); + WARN_ON(!ref && hdr->golden_frame_ts); + if (!ref) + ref = vb2_dma_contig_plane_dma_addr(&vb2_dst->vb2_buf, 0); + if (hdr->flags & V4L2_VP8_FRAME_HEADER_FLAG_SIGN_BIAS_GOLDEN) + ref |= G1_REG_ADDR_REF_TOPC_E; + vdpu_write_relaxed(vpu, ref, G1_REG_ADDR_REF(4)); + + ref = hantro_get_ref(cap_q, hdr->alt_frame_ts); + WARN_ON(!ref && hdr->alt_frame_ts); + if (!ref) + ref = vb2_dma_contig_plane_dma_addr(&vb2_dst->vb2_buf, 0); + if (hdr->flags & V4L2_VP8_FRAME_HEADER_FLAG_SIGN_BIAS_ALT) + ref |= G1_REG_ADDR_REF_TOPC_E; + vdpu_write_relaxed(vpu, ref, G1_REG_ADDR_REF(5)); +} + +static void cfg_buffers(struct hantro_ctx *ctx, + const struct v4l2_ctrl_vp8_frame_header *hdr) +{ + const struct v4l2_vp8_segment_header *seg = &hdr->segment_header; + struct hantro_dev *vpu = ctx->dev; + struct vb2_v4l2_buffer *vb2_dst; + dma_addr_t dst_dma; + u32 reg; + + vb2_dst = hantro_get_dst_buf(ctx); + + /* Set probability table buffer address */ + vdpu_write_relaxed(vpu, ctx->vp8_dec.prob_tbl.dma, + G1_REG_ADDR_QTABLE); + + /* Set segment map address */ + reg = G1_REG_FWD_PIC1_SEGMENT_BASE(ctx->vp8_dec.segment_map.dma); + if (seg->flags & V4L2_VP8_SEGMENT_HEADER_FLAG_ENABLED) { + reg |= G1_REG_FWD_PIC1_SEGMENT_E; + if (seg->flags & V4L2_VP8_SEGMENT_HEADER_FLAG_UPDATE_MAP) + reg |= G1_REG_FWD_PIC1_SEGMENT_UPD_E; + } + vdpu_write_relaxed(vpu, reg, G1_REG_FWD_PIC(0)); + + dst_dma = vb2_dma_contig_plane_dma_addr(&vb2_dst->vb2_buf, 0); + vdpu_write_relaxed(vpu, dst_dma, G1_REG_ADDR_DST); +} + +void hantro_g1_vp8_dec_run(struct hantro_ctx *ctx) +{ + const struct v4l2_ctrl_vp8_frame_header *hdr; + struct hantro_dev *vpu = ctx->dev; + size_t height = ctx->dst_fmt.height; + size_t width = ctx->dst_fmt.width; + u32 mb_width, mb_height; + u32 reg; + + hantro_prepare_run(ctx); + + hdr = hantro_get_ctrl(ctx, V4L2_CID_MPEG_VIDEO_VP8_FRAME_HEADER); + if (WARN_ON(!hdr)) + return; + + /* Reset segment_map buffer in keyframe */ + if (VP8_FRAME_IS_KEY_FRAME(hdr) && ctx->vp8_dec.segment_map.cpu) + memset(ctx->vp8_dec.segment_map.cpu, 0, + ctx->vp8_dec.segment_map.size); + + hantro_vp8_prob_update(ctx, hdr); + + reg = G1_REG_CONFIG_DEC_TIMEOUT_E | + G1_REG_CONFIG_DEC_STRENDIAN_E | + G1_REG_CONFIG_DEC_INSWAP32_E | + G1_REG_CONFIG_DEC_STRSWAP32_E | + G1_REG_CONFIG_DEC_OUTSWAP32_E | + G1_REG_CONFIG_DEC_CLK_GATE_E | + G1_REG_CONFIG_DEC_IN_ENDIAN | + G1_REG_CONFIG_DEC_OUT_ENDIAN | + G1_REG_CONFIG_DEC_MAX_BURST(16); + vdpu_write_relaxed(vpu, reg, G1_REG_CONFIG); + + reg = G1_REG_DEC_CTRL0_DEC_MODE(10); + if (!VP8_FRAME_IS_KEY_FRAME(hdr)) + reg |= G1_REG_DEC_CTRL0_PIC_INTER_E; + if (!(hdr->flags & V4L2_VP8_FRAME_HEADER_FLAG_MB_NO_SKIP_COEFF)) + reg |= G1_REG_DEC_CTRL0_SKIP_MODE; + if (hdr->lf_header.level == 0) + reg |= G1_REG_DEC_CTRL0_FILTERING_DIS; + vdpu_write_relaxed(vpu, reg, G1_REG_DEC_CTRL0); + + /* Frame dimensions */ + mb_width = VP8_MB_WIDTH(width); + mb_height = VP8_MB_HEIGHT(height); + reg = G1_REG_DEC_CTRL1_PIC_MB_WIDTH(mb_width) | + G1_REG_DEC_CTRL1_PIC_MB_HEIGHT_P(mb_height) | + G1_REG_DEC_CTRL1_PIC_MB_W_EXT(mb_width >> 9) | + G1_REG_DEC_CTRL1_PIC_MB_H_EXT(mb_height >> 8); + vdpu_write_relaxed(vpu, reg, G1_REG_DEC_CTRL1); + + /* Boolean decoder */ + reg = G1_REG_DEC_CTRL2_BOOLEAN_RANGE(hdr->coder_state.range) + | G1_REG_DEC_CTRL2_BOOLEAN_VALUE(hdr->coder_state.value); + vdpu_write_relaxed(vpu, reg, G1_REG_DEC_CTRL2); + + reg = 0; + if (hdr->version != 3) + reg |= G1_REG_DEC_CTRL4_VC1_HEIGHT_EXT; + if (hdr->version & 0x3) + reg |= G1_REG_DEC_CTRL4_BILIN_MC_E; + vdpu_write_relaxed(vpu, reg, G1_REG_DEC_CTRL4); + + cfg_lf(ctx, hdr); + cfg_qp(ctx, hdr); + cfg_parts(ctx, hdr); + cfg_tap(ctx, hdr); + cfg_ref(ctx, hdr); + cfg_buffers(ctx, hdr); + + hantro_finish_run(ctx); + + vdpu_write(vpu, G1_REG_INTERRUPT_DEC_E, G1_REG_INTERRUPT); +} diff --git a/drivers/staging/media/hantro/hantro_h1_jpeg_enc.c b/drivers/staging/media/hantro/hantro_h1_jpeg_enc.c index 0c1e3043dc7e..ecd34a7db190 100644 --- a/drivers/staging/media/hantro/hantro_h1_jpeg_enc.c +++ b/drivers/staging/media/hantro/hantro_h1_jpeg_enc.c @@ -84,8 +84,10 @@ void hantro_h1_jpeg_enc_run(struct hantro_ctx *ctx) struct hantro_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); + src_buf = hantro_get_src_buf(ctx); + dst_buf = hantro_get_dst_buf(ctx); + + hantro_prepare_run(ctx); memset(&jpeg_ctx, 0, sizeof(jpeg_ctx)); jpeg_ctx.buffer = vb2_plane_vaddr(&dst_buf->vb2_buf, 0); @@ -119,7 +121,8 @@ void hantro_h1_jpeg_enc_run(struct hantro_ctx *ctx) | H1_REG_ENC_CTRL_ENC_MODE_JPEG | H1_REG_ENC_PIC_INTRA | H1_REG_ENC_CTRL_EN_BIT; - /* Kick the watchdog and start encoding */ - schedule_delayed_work(&vpu->watchdog_work, msecs_to_jiffies(2000)); + + hantro_finish_run(ctx); + vepu_write(vpu, reg, H1_REG_ENC_CTRL); } diff --git a/drivers/staging/media/hantro/hantro_h264.c b/drivers/staging/media/hantro/hantro_h264.c new file mode 100644 index 000000000000..0d758e0c0f99 --- /dev/null +++ b/drivers/staging/media/hantro/hantro_h264.c @@ -0,0 +1,646 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Rockchip RK3288 VPU codec driver + * + * Copyright (c) 2014 Rockchip Electronics Co., Ltd. + * Hertz Wong <hertz.wong@rock-chips.com> + * Herman Chen <herman.chen@rock-chips.com> + * + * Copyright (C) 2014 Google, Inc. + * Tomasz Figa <tfiga@chromium.org> + */ + +#include <linux/types.h> +#include <linux/sort.h> +#include <media/v4l2-mem2mem.h> + +#include "hantro.h" +#include "hantro_hw.h" + +/* Size with u32 units. */ +#define CABAC_INIT_BUFFER_SIZE (460 * 2) +#define POC_BUFFER_SIZE 34 +#define SCALING_LIST_SIZE (6 * 16 + 6 * 64) + +#define POC_CMP(p0, p1) ((p0) < (p1) ? -1 : 1) + +/* Data structure describing auxiliary buffer format. */ +struct hantro_h264_dec_priv_tbl { + u32 cabac_table[CABAC_INIT_BUFFER_SIZE]; + u32 poc[POC_BUFFER_SIZE]; + u8 scaling_list[SCALING_LIST_SIZE]; +}; + +/* + * Constant CABAC table. + * From drivers/media/platform/rk3288-vpu/rk3288_vpu_hw_h264d.c + * in https://chromium.googlesource.com/chromiumos/third_party/kernel, + * chromeos-3.14 branch. + */ +static const u32 h264_cabac_table[] = { + 0x14f10236, 0x034a14f1, 0x0236034a, 0xe47fe968, 0xfa35ff36, 0x07330000, + 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, + 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, + 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, + 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, + 0x0029003f, 0x003f003f, 0xf7530456, 0x0061f948, 0x0d29033e, 0x000b0137, + 0x0045ef7f, 0xf3660052, 0xf94aeb6b, 0xe57fe17f, 0xe87fee5f, 0xe57feb72, + 0xe27fef7b, 0xf473f07a, 0xf573f43f, 0xfe44f154, 0xf368fd46, 0xf85df65a, + 0xe27fff4a, 0xfa61f95b, 0xec7ffc38, 0xfb52f94c, 0xea7df95d, 0xf557fd4d, + 0xfb47fc3f, 0xfc44f454, 0xf93ef941, 0x083d0538, 0xfe420140, 0x003dfe4e, + 0x01320734, 0x0a23002c, 0x0b26012d, 0x002e052c, 0x1f110133, 0x07321c13, + 0x10210e3e, 0xf36cf164, 0xf365f35b, 0xf45ef658, 0xf054f656, 0xf953f357, + 0xed5e0146, 0x0048fb4a, 0x123bf866, 0xf164005f, 0xfc4b0248, 0xf54bfd47, + 0x0f2ef345, 0x003e0041, 0x1525f148, 0x09391036, 0x003e0c48, 0x18000f09, + 0x08190d12, 0x0f090d13, 0x0a250c12, 0x061d1421, 0x0f1e042d, 0x013a003e, + 0x073d0c26, 0x0b2d0f27, 0x0b2a0d2c, 0x102d0c29, 0x0a311e22, 0x122a0a37, + 0x1133112e, 0x00591aed, 0x16ef1aef, 0x1ee71cec, 0x21e925e5, 0x21e928e4, + 0x26ef21f5, 0x28f129fa, 0x26012911, 0x1efa1b03, 0x1a1625f0, 0x23fc26f8, + 0x26fd2503, 0x26052a00, 0x23102716, 0x0e301b25, 0x153c0c44, 0x0261fd47, + 0xfa2afb32, 0xfd36fe3e, 0x003a013f, 0xfe48ff4a, 0xf75bfb43, 0xfb1bfd27, + 0xfe2c002e, 0xf040f844, 0xf64efa4d, 0xf656f45c, 0xf137f63c, 0xfa3efc41, + 0xf449f84c, 0xf950f758, 0xef6ef561, 0xec54f54f, 0xfa49fc4a, 0xf356f360, + 0xf561ed75, 0xf84efb21, 0xfc30fe35, 0xfd3ef347, 0xf64ff456, 0xf35af261, + 0x0000fa5d, 0xfa54f84f, 0x0042ff47, 0x003efe3c, 0xfe3bfb4b, 0xfd3efc3a, + 0xf742ff4f, 0x00470344, 0x0a2cf93e, 0x0f240e28, 0x101b0c1d, 0x012c1424, + 0x1220052a, 0x01300a3e, 0x112e0940, 0xf468f561, 0xf060f958, 0xf855f955, + 0xf755f358, 0x0442fd4d, 0xfd4cfa4c, 0x0a3aff4c, 0xff53f963, 0xf25f025f, + 0x004cfb4a, 0x0046f54b, 0x01440041, 0xf249033e, 0x043eff44, 0xf34b0b37, + 0x05400c46, 0x0f060613, 0x07100c0e, 0x120d0d0b, 0x0d0f0f10, 0x0c170d17, + 0x0f140e1a, 0x0e2c1128, 0x112f1811, 0x15151916, 0x1f1b161d, 0x13230e32, + 0x0a39073f, 0xfe4dfc52, 0xfd5e0945, 0xf46d24dd, 0x24de20e6, 0x25e22ce0, + 0x22ee22f1, 0x28f121f9, 0x23fb2100, 0x2602210d, 0x17230d3a, 0x1dfd1a00, + 0x161e1ff9, 0x23f122fd, 0x220324ff, 0x2205200b, 0x2305220c, 0x270b1e1d, + 0x221a1d27, 0x13421f15, 0x1f1f1932, 0xef78ec70, 0xee72f555, 0xf15cf259, + 0xe647f151, 0xf2500044, 0xf246e838, 0xe944e832, 0xf54a17f3, 0x1af328f1, + 0x31f22c03, 0x2d062c22, 0x21361352, 0xfd4bff17, 0x0122012b, 0x0036fe37, + 0x003d0140, 0x0044f75c, 0xf26af361, 0xf15af45a, 0xee58f649, 0xf74ff256, + 0xf649f646, 0xf645fb42, 0xf740fb3a, 0x023b15f6, 0x18f51cf8, 0x1cff1d03, + 0x1d092314, 0x1d240e43, 0x14f10236, 0x034a14f1, 0x0236034a, 0xe47fe968, + 0xfa35ff36, 0x07331721, 0x17021500, 0x01090031, 0xdb760539, 0xf34ef541, + 0x013e0c31, 0xfc491132, 0x1240092b, 0x1d001a43, 0x105a0968, 0xd27fec68, + 0x0143f34e, 0xf541013e, 0xfa56ef5f, 0xfa3d092d, 0xfd45fa51, 0xf5600637, + 0x0743fb56, 0x0258003a, 0xfd4cf65e, 0x05360445, 0xfd510058, 0xf943fb4a, + 0xfc4afb50, 0xf948013a, 0x0029003f, 0x003f003f, 0xf7530456, 0x0061f948, + 0x0d29033e, 0x002dfc4e, 0xfd60e57e, 0xe462e765, 0xe943e452, 0xec5ef053, + 0xea6eeb5b, 0xee66f35d, 0xe37ff95c, 0xfb59f960, 0xf36cfd2e, 0xff41ff39, + 0xf75dfd4a, 0xf75cf857, 0xe97e0536, 0x063c063b, 0x0645ff30, 0x0044fc45, + 0xf858fe55, 0xfa4eff4b, 0xf94d0236, 0x0532fd44, 0x0132062a, 0xfc51013f, + 0xfc460043, 0x0239fe4c, 0x0b230440, 0x013d0b23, 0x12190c18, 0x0d1d0d24, + 0xf65df949, 0xfe490d2e, 0x0931f964, 0x09350235, 0x0535fe3d, 0x00380038, + 0xf33ffb3c, 0xff3e0439, 0xfa450439, 0x0e270433, 0x0d440340, 0x013d093f, + 0x07321027, 0x052c0434, 0x0b30fb3c, 0xff3b003b, 0x1621052c, 0x0e2bff4e, + 0x003c0945, 0x0b1c0228, 0x032c0031, 0x002e022c, 0x0233002f, 0x0427023e, + 0x062e0036, 0x0336023a, 0x043f0633, 0x06390735, 0x06340637, 0x0b2d0e24, + 0x0835ff52, 0x0737fd4e, 0x0f2e161f, 0xff541907, 0x1ef91c03, 0x1c042000, + 0x22ff1e06, 0x1e062009, 0x1f131a1b, 0x1a1e2514, 0x1c221146, 0x0143053b, + 0x0943101e, 0x12201223, 0x161d181f, 0x1726122b, 0x14290b3f, 0x093b0940, + 0xff5efe59, 0xf76cfa4c, 0xfe2c002d, 0x0034fd40, 0xfe3bfc46, 0xfc4bf852, + 0xef66f74d, 0x0318002a, 0x00300037, 0xfa3bf947, 0xf453f557, 0xe277013a, + 0xfd1dff24, 0x0126022b, 0xfa37003a, 0x0040fd4a, 0xf65a0046, 0xfc1d051f, + 0x072a013b, 0xfe3afd48, 0xfd51f561, 0x003a0805, 0x0a0e0e12, 0x0d1b0228, + 0x003afd46, 0xfa4ff855, 0x0000f36a, 0xf06af657, 0xeb72ee6e, 0xf262ea6e, + 0xeb6aee67, 0xeb6be96c, 0xe670f660, 0xf45ffb5b, 0xf75dea5e, 0xfb560943, + 0xfc50f655, 0xff46073c, 0x093a053d, 0x0c320f32, 0x12311136, 0x0a29072e, + 0xff330731, 0x08340929, 0x062f0237, 0x0d290a2c, 0x06320535, 0x0d31043f, + 0x0640fe45, 0xfe3b0646, 0x0a2c091f, 0x0c2b0335, 0x0e220a26, 0xfd340d28, + 0x1120072c, 0x07260d32, 0x0a391a2b, 0x0e0b0b0e, 0x090b120b, 0x150917fe, + 0x20f120f1, 0x22eb27e9, 0x2adf29e1, 0x2ee426f4, 0x151d2de8, 0x35d330e6, + 0x41d52bed, 0x27f61e09, 0x121a141b, 0x0039f252, 0xfb4bed61, 0xdd7d1b00, + 0x1c001ffc, 0x1b062208, 0x1e0a1816, 0x21131620, 0x1a1f1529, 0x1a2c172f, + 0x10410e47, 0x083c063f, 0x11411518, 0x17141a17, 0x1b201c17, 0x1c181728, + 0x18201c1d, 0x172a1339, 0x1635163d, 0x0b560c28, 0x0b330e3b, 0xfc4ff947, + 0xfb45f746, 0xf842f644, 0xed49f445, 0xf046f143, 0xec3eed46, 0xf042ea41, + 0xec3f09fe, 0x1af721f7, 0x27f929fe, 0x2d033109, 0x2d1b243b, 0xfa42f923, + 0xf92af82d, 0xfb30f438, 0xfa3cfb3e, 0xf842f84c, 0xfb55fa51, 0xf64df951, + 0xef50ee49, 0xfc4af653, 0xf747f743, 0xff3df842, 0xf242003b, 0x023b15f3, + 0x21f227f9, 0x2efe3302, 0x3c063d11, 0x37222a3e, 0x14f10236, 0x034a14f1, + 0x0236034a, 0xe47fe968, 0xfa35ff36, 0x07331619, 0x22001000, 0xfe090429, + 0xe3760241, 0xfa47f34f, 0x05340932, 0xfd460a36, 0x1a221316, 0x28003902, + 0x29241a45, 0xd37ff165, 0xfc4cfa47, 0xf34f0534, 0x0645f35a, 0x0034082b, + 0xfe45fb52, 0xf660023b, 0x024bfd57, 0xfd640138, 0xfd4afa55, 0x003bfd51, + 0xf956fb5f, 0xff42ff4d, 0x0146fe56, 0xfb48003d, 0x0029003f, 0x003f003f, + 0xf7530456, 0x0061f948, 0x0d29033e, 0x0d0f0733, 0x0250d97f, 0xee5bef60, + 0xe651dd62, 0xe866e961, 0xe577e863, 0xeb6eee66, 0xdc7f0050, 0xfb59f95e, + 0xfc5c0027, 0x0041f154, 0xdd7ffe49, 0xf468f75b, 0xe17f0337, 0x07380737, + 0x083dfd35, 0x0044f94a, 0xf758f367, 0xf35bf759, 0xf25cf84c, 0xf457e96e, + 0xe869f64e, 0xec70ef63, 0xb27fba7f, 0xce7fd27f, 0xfc42fb4e, 0xfc47f848, + 0x023bff37, 0xf946fa4b, 0xf859de77, 0xfd4b2014, 0x1e16d47f, 0x0036fb3d, + 0x003aff3c, 0xfd3df843, 0xe754f24a, 0xfb410534, 0x0239003d, 0xf745f546, + 0x1237fc47, 0x003a073d, 0x09291219, 0x0920052b, 0x092f002c, 0x0033022e, + 0x1326fc42, 0x0f260c2a, 0x09220059, 0x042d0a1c, 0x0a1f21f5, 0x34d5120f, + 0x1c0023ea, 0x26e72200, 0x27ee20f4, 0x66a20000, 0x38f121fc, 0x1d0a25fb, + 0x33e327f7, 0x34de45c6, 0x43c12cfb, 0x200737e3, 0x20010000, 0x1b2421e7, + 0x22e224e4, 0x26e426e5, 0x22ee23f0, 0x22f220f8, 0x25fa2300, 0x1e0a1c12, + 0x1a191d29, 0x004b0248, 0x084d0e23, 0x121f1123, 0x151e112d, 0x142a122d, + 0x1b1a1036, 0x07421038, 0x0b490a43, 0xf674e970, 0xf147f93d, 0x0035fb42, + 0xf54df750, 0xf754f657, 0xde7feb65, 0xfd27fb35, 0xf93df54b, 0xf14def5b, + 0xe76be76f, 0xe47af54c, 0xf62cf634, 0xf639f73a, 0xf048f945, 0xfc45fb4a, + 0xf7560242, 0xf7220120, 0x0b1f0534, 0xfe37fe43, 0x0049f859, 0x03340704, + 0x0a081108, 0x10130325, 0xff3dfb49, 0xff46fc4e, 0x0000eb7e, 0xe97cec6e, + 0xe67ee77c, 0xef69e579, 0xe575ef66, 0xe675e574, 0xdf7af65f, 0xf264f85f, + 0xef6fe472, 0xfa59fe50, 0xfc52f755, 0xf851ff48, 0x05400143, 0x09380045, + 0x01450745, 0xf945fa43, 0xf04dfe40, 0x023dfa43, 0xfd400239, 0xfd41fd42, + 0x003e0933, 0xff42fe47, 0xfe4bff46, 0xf7480e3c, 0x1025002f, 0x12230b25, + 0x0c290a29, 0x02300c29, 0x0d29003b, 0x03321328, 0x03421232, 0x13fa12fa, + 0x0e001af4, 0x1ff021e7, 0x21ea25e4, 0x27e22ae2, 0x2fd62ddc, 0x31de29ef, + 0x200945b9, 0x3fc142c0, 0x4db636d9, 0x34dd29f6, 0x240028ff, 0x1e0e1c1a, + 0x17250c37, 0x0b4125df, 0x27dc28db, 0x26e22edf, 0x2ae228e8, 0x31e326f4, + 0x28f626fd, 0x2efb1f14, 0x1d1e192c, 0x0c300b31, 0x1a2d1616, 0x17161b15, + 0x21141a1c, 0x1e181b22, 0x122a1927, 0x12320c46, 0x15360e47, 0x0b531920, + 0x15311536, 0xfb55fa51, 0xf64df951, 0xef50ee49, 0xfc4af653, 0xf747f743, + 0xff3df842, 0xf242003b, 0x023b11f6, 0x20f32af7, 0x31fb3500, 0x4003440a, + 0x421b2f39, 0xfb470018, 0xff24fe2a, 0xfe34f739, 0xfa3ffc41, 0xfc43f952, + 0xfd51fd4c, 0xf948fa4e, 0xf448f244, 0xfd46fa4c, 0xfb42fb3e, 0x0039fc3d, + 0xf73c0136, 0x023a11f6, 0x20f32af7, 0x31fb3500, 0x4003440a, 0x421b2f39, + 0x14f10236, 0x034a14f1, 0x0236034a, 0xe47fe968, 0xfa35ff36, 0x07331d10, + 0x19000e00, 0xf633fd3e, 0xe5631a10, 0xfc55e866, 0x05390639, 0xef490e39, + 0x1428140a, 0x1d003600, 0x252a0c61, 0xe07fea75, 0xfe4afc55, 0xe8660539, + 0xfa5df258, 0xfa2c0437, 0xf559f167, 0xeb741339, 0x143a0454, 0x0660013f, + 0xfb55f36a, 0x053f064b, 0xfd5aff65, 0x0337fc4f, 0xfe4bf461, 0xf932013c, + 0x0029003f, 0x003f003f, 0xf7530456, 0x0061f948, 0x0d29033e, 0x0722f758, + 0xec7fdc7f, 0xef5bf25f, 0xe754e756, 0xf459ef5b, 0xe17ff24c, 0xee67f35a, + 0xdb7f0b50, 0x054c0254, 0x054efa37, 0x043df253, 0xdb7ffb4f, 0xf568f55b, + 0xe27f0041, 0xfe4f0048, 0xfc5cfa38, 0x0344f847, 0xf362fc56, 0xf458fb52, + 0xfd48fc43, 0xf848f059, 0xf745ff3b, 0x05420439, 0xfc47fe47, 0x023aff4a, + 0xfc2cff45, 0x003ef933, 0xfc2ffa2a, 0xfd29fa35, 0x084cf74e, 0xf5530934, + 0x0043fb5a, 0x0143f148, 0xfb4bf850, 0xeb53eb40, 0xf31fe740, 0xe35e094b, + 0x113ff84a, 0xfb23fe1b, 0x0d5b0341, 0xf945084d, 0xf642033e, 0xfd44ec51, + 0x001e0107, 0xfd17eb4a, 0x1042e97c, 0x11252cee, 0x32deea7f, 0x0427002a, + 0x07220b1d, 0x081f0625, 0x072a0328, 0x08210d2b, 0x0d24042f, 0x0337023a, + 0x063c082c, 0x0b2c0e2a, 0x07300438, 0x04340d25, 0x0931133a, 0x0a300c2d, + 0x00451421, 0x083f23ee, 0x21e71cfd, 0x180a1b00, 0x22f234d4, 0x27e81311, + 0x1f19241d, 0x1821220f, 0x1e141649, 0x1422131f, 0x1b2c1310, 0x0f240f24, + 0x151c1915, 0x1e141f0c, 0x1b10182a, 0x005d0e38, 0x0f391a26, 0xe87fe873, + 0xea52f73e, 0x0035003b, 0xf255f359, 0xf35ef55c, 0xe37feb64, 0xf239f443, + 0xf547f64d, 0xeb55f058, 0xe968f162, 0xdb7ff652, 0xf830f83d, 0xf842f946, + 0xf24bf64f, 0xf753f45c, 0xee6cfc4f, 0xea45f04b, 0xfe3a013a, 0xf34ef753, + 0xfc51f363, 0xf351fa26, 0xf33efa3a, 0xfe3bf049, 0xf64cf356, 0xf753f657, + 0x0000ea7f, 0xe77fe778, 0xe57fed72, 0xe975e776, 0xe675e871, 0xe476e178, + 0xdb7cf65e, 0xf166f663, 0xf36ace7f, 0xfb5c1139, 0xfb56f35e, 0xf45bfe4d, + 0x0047ff49, 0x0440f951, 0x05400f39, 0x01430044, 0xf6430144, 0x004d0240, + 0x0044fb4e, 0x0737053b, 0x02410e36, 0x0f2c053c, 0x0246fe4c, 0xee560c46, + 0x0540f446, 0x0b370538, 0x00450241, 0xfa4a0536, 0x0736fa4c, 0xf552fe4d, + 0xfe4d192a, 0x11f310f7, 0x11f41beb, 0x25e229d8, 0x2ad730d1, 0x27e02ed8, + 0x34cd2ed7, 0x34d92bed, 0x200b3dc9, 0x38d23ece, 0x51bd2dec, 0x23fe1c0f, + 0x22012701, 0x1e111426, 0x122d0f36, 0x004f24f0, 0x25f225ef, 0x2001220f, + 0x1d0f1819, 0x22161f10, 0x23121f1c, 0x2129241c, 0x1b2f153e, 0x121f131a, + 0x24181817, 0x1b10181e, 0x1f1d1629, 0x162a103c, 0x0f340e3c, 0x034ef07b, + 0x15351638, 0x193d1521, 0x1332113d, 0xfd4ef84a, 0xf748f648, 0xee4bf447, + 0xf53ffb46, 0xef4bf248, 0xf043f835, 0xf23bf734, 0xf54409fe, 0x1ef61ffc, + 0x21ff2107, 0x1f0c2517, 0x1f261440, 0xf747f925, 0xf82cf531, 0xf638f43b, + 0xf83ff743, 0xfa44f64f, 0xfd4ef84a, 0xf748f648, 0xee4bf447, 0xf53ffb46, + 0xef4bf248, 0xf043f835, 0xf23bf734, 0xf54409fe, 0x1ef61ffc, 0x21ff2107, + 0x1f0c2517, 0x1f261440 +}; + +/* + * NOTE: The scaling lists are in zig-zag order, apply inverse scanning process + * to get the values in matrix order. In addition, the hardware requires bytes + * swapped within each subsequent 4 bytes. Both arrays below include both + * transformations. + */ +static const u32 zig_zag_4x4[] = { + 3, 2, 7, 11, 6, 1, 0, 5, 10, 15, 14, 9, 4, 8, 13, 12 +}; + +static const u32 zig_zag_8x8[] = { + 3, 2, 11, 19, 10, 1, 0, 9, 18, 27, 35, 26, 17, 8, 7, 6, + 15, 16, 25, 34, 43, 51, 42, 33, 24, 23, 14, 5, 4, 13, 22, 31, + 32, 41, 50, 59, 58, 49, 40, 39, 30, 21, 12, 20, 29, 38, 47, 48, + 57, 56, 55, 46, 37, 28, 36, 45, 54, 63, 62, 53, 44, 52, 61, 60 +}; + +static void +reorder_scaling_list(struct hantro_ctx *ctx) +{ + const struct hantro_h264_dec_ctrls *ctrls = &ctx->h264_dec.ctrls; + const struct v4l2_ctrl_h264_scaling_matrix *scaling = ctrls->scaling; + const size_t num_list_4x4 = ARRAY_SIZE(scaling->scaling_list_4x4); + const size_t list_len_4x4 = ARRAY_SIZE(scaling->scaling_list_4x4[0]); + const size_t num_list_8x8 = ARRAY_SIZE(scaling->scaling_list_8x8); + const size_t list_len_8x8 = ARRAY_SIZE(scaling->scaling_list_8x8[0]); + struct hantro_h264_dec_priv_tbl *tbl = ctx->h264_dec.priv.cpu; + u8 *dst = tbl->scaling_list; + const u8 *src; + int i, j; + + BUILD_BUG_ON(ARRAY_SIZE(zig_zag_4x4) != list_len_4x4); + BUILD_BUG_ON(ARRAY_SIZE(zig_zag_8x8) != list_len_8x8); + BUILD_BUG_ON(ARRAY_SIZE(tbl->scaling_list) != + num_list_4x4 * list_len_4x4 + + num_list_8x8 * list_len_8x8); + + src = &scaling->scaling_list_4x4[0][0]; + for (i = 0; i < num_list_4x4; ++i) { + for (j = 0; j < list_len_4x4; ++j) + dst[zig_zag_4x4[j]] = src[j]; + src += list_len_4x4; + dst += list_len_4x4; + } + + src = &scaling->scaling_list_8x8[0][0]; + for (i = 0; i < num_list_8x8; ++i) { + for (j = 0; j < list_len_8x8; ++j) + dst[zig_zag_8x8[j]] = src[j]; + src += list_len_8x8; + dst += list_len_8x8; + } +} + +static void prepare_table(struct hantro_ctx *ctx) +{ + const struct hantro_h264_dec_ctrls *ctrls = &ctx->h264_dec.ctrls; + const struct v4l2_ctrl_h264_decode_params *dec_param = ctrls->decode; + struct hantro_h264_dec_priv_tbl *tbl = ctx->h264_dec.priv.cpu; + const struct v4l2_h264_dpb_entry *dpb = ctx->h264_dec.dpb; + int i; + + for (i = 0; i < HANTRO_H264_DPB_SIZE; ++i) { + tbl->poc[i * 2] = dpb[i].top_field_order_cnt; + tbl->poc[i * 2 + 1] = dpb[i].bottom_field_order_cnt; + } + + tbl->poc[32] = dec_param->top_field_order_cnt; + tbl->poc[33] = dec_param->bottom_field_order_cnt; + + reorder_scaling_list(ctx); +} + +struct hantro_h264_reflist_builder { + const struct v4l2_h264_dpb_entry *dpb; + s32 pocs[HANTRO_H264_DPB_SIZE]; + u8 unordered_reflist[HANTRO_H264_DPB_SIZE]; + s32 curpoc; + u8 num_valid; +}; + +static s32 get_poc(enum v4l2_field field, s32 top_field_order_cnt, + s32 bottom_field_order_cnt) +{ + switch (field) { + case V4L2_FIELD_TOP: + return top_field_order_cnt; + case V4L2_FIELD_BOTTOM: + return bottom_field_order_cnt; + default: + break; + } + + return min(top_field_order_cnt, bottom_field_order_cnt); +} + +static void +init_reflist_builder(struct hantro_ctx *ctx, + struct hantro_h264_reflist_builder *b) +{ + const struct v4l2_ctrl_h264_decode_params *dec_param; + struct vb2_v4l2_buffer *buf = hantro_get_dst_buf(ctx); + const struct v4l2_h264_dpb_entry *dpb = ctx->h264_dec.dpb; + struct vb2_queue *cap_q = &ctx->fh.m2m_ctx->cap_q_ctx.q; + unsigned int i; + + dec_param = ctx->h264_dec.ctrls.decode; + + memset(b, 0, sizeof(*b)); + b->dpb = dpb; + b->curpoc = get_poc(buf->field, dec_param->top_field_order_cnt, + dec_param->bottom_field_order_cnt); + + for (i = 0; i < ARRAY_SIZE(ctx->h264_dec.dpb); i++) { + int buf_idx; + + if (!(dpb[i].flags & V4L2_H264_DPB_ENTRY_FLAG_ACTIVE)) + continue; + + buf_idx = vb2_find_timestamp(cap_q, dpb[i].reference_ts, 0); + if (buf_idx < 0) + continue; + + buf = to_vb2_v4l2_buffer(vb2_get_buffer(cap_q, buf_idx)); + b->pocs[i] = get_poc(buf->field, dpb[i].top_field_order_cnt, + dpb[i].bottom_field_order_cnt); + b->unordered_reflist[b->num_valid] = i; + b->num_valid++; + } + + for (i = b->num_valid; i < ARRAY_SIZE(ctx->h264_dec.dpb); i++) + b->unordered_reflist[i] = i; +} + +static int p_ref_list_cmp(const void *ptra, const void *ptrb, const void *data) +{ + const struct hantro_h264_reflist_builder *builder = data; + const struct v4l2_h264_dpb_entry *a, *b; + u8 idxa, idxb; + + idxa = *((u8 *)ptra); + idxb = *((u8 *)ptrb); + a = &builder->dpb[idxa]; + b = &builder->dpb[idxb]; + + if ((a->flags & V4L2_H264_DPB_ENTRY_FLAG_LONG_TERM) != + (b->flags & V4L2_H264_DPB_ENTRY_FLAG_LONG_TERM)) { + /* Short term pics firt. */ + if (!(a->flags & V4L2_H264_DPB_ENTRY_FLAG_LONG_TERM)) + return -1; + else + return 1; + } + + /* + * Short term pics in descending pic num order, long term ones in + * ascending order. + */ + if (!(a->flags & V4L2_H264_DPB_ENTRY_FLAG_LONG_TERM)) + return b->frame_num - a->frame_num; + + return a->pic_num - b->pic_num; +} + +static int b0_ref_list_cmp(const void *ptra, const void *ptrb, const void *data) +{ + const struct hantro_h264_reflist_builder *builder = data; + const struct v4l2_h264_dpb_entry *a, *b; + s32 poca, pocb; + u8 idxa, idxb; + + idxa = *((u8 *)ptra); + idxb = *((u8 *)ptrb); + a = &builder->dpb[idxa]; + b = &builder->dpb[idxb]; + + if ((a->flags & V4L2_H264_DPB_ENTRY_FLAG_LONG_TERM) != + (b->flags & V4L2_H264_DPB_ENTRY_FLAG_LONG_TERM)) { + /* Short term pics firt. */ + if (!(a->flags & V4L2_H264_DPB_ENTRY_FLAG_LONG_TERM)) + return -1; + else + return 1; + } + + /* Long term pics in ascending pic num order. */ + if (a->flags & V4L2_H264_DPB_ENTRY_FLAG_LONG_TERM) + return a->pic_num - b->pic_num; + + poca = builder->pocs[idxa]; + pocb = builder->pocs[idxb]; + + /* + * Short term pics with POC < cur POC first in POC descending order + * followed by short term pics with POC > cur POC in POC ascending + * order. + */ + if ((poca < builder->curpoc) != (pocb < builder->curpoc)) + return POC_CMP(poca, pocb); + else if (poca < builder->curpoc) + return POC_CMP(pocb, poca); + + return POC_CMP(poca, pocb); +} + +static int b1_ref_list_cmp(const void *ptra, const void *ptrb, const void *data) +{ + const struct hantro_h264_reflist_builder *builder = data; + const struct v4l2_h264_dpb_entry *a, *b; + s32 poca, pocb; + u8 idxa, idxb; + + idxa = *((u8 *)ptra); + idxb = *((u8 *)ptrb); + a = &builder->dpb[idxa]; + b = &builder->dpb[idxb]; + + if ((a->flags & V4L2_H264_DPB_ENTRY_FLAG_LONG_TERM) != + (b->flags & V4L2_H264_DPB_ENTRY_FLAG_LONG_TERM)) { + /* Short term pics firt. */ + if (!(a->flags & V4L2_H264_DPB_ENTRY_FLAG_LONG_TERM)) + return -1; + else + return 1; + } + + /* Long term pics in ascending pic num order. */ + if (a->flags & V4L2_H264_DPB_ENTRY_FLAG_LONG_TERM) + return a->pic_num - b->pic_num; + + poca = builder->pocs[idxa]; + pocb = builder->pocs[idxb]; + + /* + * Short term pics with POC > cur POC first in POC ascending order + * followed by short term pics with POC > cur POC in POC descending + * order. + */ + if ((poca < builder->curpoc) != (pocb < builder->curpoc)) + return POC_CMP(pocb, poca); + else if (poca < builder->curpoc) + return POC_CMP(pocb, poca); + + return POC_CMP(poca, pocb); +} + +static void +build_p_ref_list(const struct hantro_h264_reflist_builder *builder, + u8 *reflist) +{ + memcpy(reflist, builder->unordered_reflist, + sizeof(builder->unordered_reflist)); + sort_r(reflist, builder->num_valid, sizeof(*reflist), + p_ref_list_cmp, NULL, builder); +} + +static void +build_b_ref_lists(const struct hantro_h264_reflist_builder *builder, + u8 *b0_reflist, u8 *b1_reflist) +{ + memcpy(b0_reflist, builder->unordered_reflist, + sizeof(builder->unordered_reflist)); + sort_r(b0_reflist, builder->num_valid, sizeof(*b0_reflist), + b0_ref_list_cmp, NULL, builder); + + memcpy(b1_reflist, builder->unordered_reflist, + sizeof(builder->unordered_reflist)); + sort_r(b1_reflist, builder->num_valid, sizeof(*b1_reflist), + b1_ref_list_cmp, NULL, builder); + + if (builder->num_valid > 1 && + !memcmp(b1_reflist, b0_reflist, builder->num_valid)) + swap(b1_reflist[0], b1_reflist[1]); +} + +static bool dpb_entry_match(const struct v4l2_h264_dpb_entry *a, + const struct v4l2_h264_dpb_entry *b) +{ + return a->top_field_order_cnt == b->top_field_order_cnt && + a->bottom_field_order_cnt == b->bottom_field_order_cnt; +} + +static void update_dpb(struct hantro_ctx *ctx) +{ + const struct v4l2_ctrl_h264_decode_params *dec_param; + DECLARE_BITMAP(new, ARRAY_SIZE(dec_param->dpb)) = { 0, }; + DECLARE_BITMAP(used, ARRAY_SIZE(dec_param->dpb)) = { 0, }; + unsigned int i, j; + + dec_param = ctx->h264_dec.ctrls.decode; + + /* Disable all entries by default. */ + for (i = 0; i < ARRAY_SIZE(ctx->h264_dec.dpb); i++) + ctx->h264_dec.dpb[i].flags &= ~V4L2_H264_DPB_ENTRY_FLAG_ACTIVE; + + /* Try to match new DPB entries with existing ones by their POCs. */ + for (i = 0; i < ARRAY_SIZE(dec_param->dpb); i++) { + const struct v4l2_h264_dpb_entry *ndpb = &dec_param->dpb[i]; + + if (!(ndpb->flags & V4L2_H264_DPB_ENTRY_FLAG_ACTIVE)) + continue; + + /* + * To cut off some comparisons, iterate only on target DPB + * entries which are not used yet. + */ + for_each_clear_bit(j, used, ARRAY_SIZE(ctx->h264_dec.dpb)) { + struct v4l2_h264_dpb_entry *cdpb; + + cdpb = &ctx->h264_dec.dpb[j]; + if (cdpb->flags & V4L2_H264_DPB_ENTRY_FLAG_ACTIVE || + !dpb_entry_match(cdpb, ndpb)) + continue; + + *cdpb = *ndpb; + set_bit(j, used); + break; + } + + if (j == ARRAY_SIZE(ctx->h264_dec.dpb)) + set_bit(i, new); + } + + /* For entries that could not be matched, use remaining free slots. */ + for_each_set_bit(i, new, ARRAY_SIZE(dec_param->dpb)) { + const struct v4l2_h264_dpb_entry *ndpb = &dec_param->dpb[i]; + struct v4l2_h264_dpb_entry *cdpb; + + /* + * Both arrays are of the same sizes, so there is no way + * we can end up with no space in target array, unless + * something is buggy. + */ + j = find_first_zero_bit(used, ARRAY_SIZE(ctx->h264_dec.dpb)); + if (WARN_ON(j >= ARRAY_SIZE(ctx->h264_dec.dpb))) + return; + + cdpb = &ctx->h264_dec.dpb[j]; + *cdpb = *ndpb; + set_bit(j, used); + } +} + +struct vb2_buffer *hantro_h264_get_ref_buf(struct hantro_ctx *ctx, + unsigned int dpb_idx) +{ + struct vb2_queue *cap_q = &ctx->fh.m2m_ctx->cap_q_ctx.q; + struct v4l2_h264_dpb_entry *dpb = ctx->h264_dec.dpb; + struct vb2_buffer *buf; + int buf_idx = -1; + + if (dpb[dpb_idx].flags & V4L2_H264_DPB_ENTRY_FLAG_ACTIVE) + buf_idx = vb2_find_timestamp(cap_q, + dpb[dpb_idx].reference_ts, 0); + + if (buf_idx >= 0) { + buf = vb2_get_buffer(cap_q, buf_idx); + } else { + struct vb2_v4l2_buffer *dst_buf; + + /* + * If a DPB entry is unused or invalid, address of current + * destination buffer is returned. + */ + dst_buf = hantro_get_dst_buf(ctx); + buf = &dst_buf->vb2_buf; + } + + return buf; +} + +int hantro_h264_dec_prepare_run(struct hantro_ctx *ctx) +{ + struct hantro_h264_dec_hw_ctx *h264_ctx = &ctx->h264_dec; + struct hantro_h264_dec_ctrls *ctrls = &h264_ctx->ctrls; + struct hantro_h264_reflist_builder reflist_builder; + + hantro_prepare_run(ctx); + + ctrls->scaling = + hantro_get_ctrl(ctx, V4L2_CID_MPEG_VIDEO_H264_SCALING_MATRIX); + if (WARN_ON(!ctrls->scaling)) + return -EINVAL; + + ctrls->decode = + hantro_get_ctrl(ctx, V4L2_CID_MPEG_VIDEO_H264_DECODE_PARAMS); + if (WARN_ON(!ctrls->decode)) + return -EINVAL; + + ctrls->slices = + hantro_get_ctrl(ctx, V4L2_CID_MPEG_VIDEO_H264_SLICE_PARAMS); + if (WARN_ON(!ctrls->slices)) + return -EINVAL; + + ctrls->sps = + hantro_get_ctrl(ctx, V4L2_CID_MPEG_VIDEO_H264_SPS); + if (WARN_ON(!ctrls->sps)) + return -EINVAL; + + ctrls->pps = + hantro_get_ctrl(ctx, V4L2_CID_MPEG_VIDEO_H264_PPS); + if (WARN_ON(!ctrls->pps)) + return -EINVAL; + + /* Update the DPB with new refs. */ + update_dpb(ctx); + + /* Prepare data in memory. */ + prepare_table(ctx); + + /* Build the P/B{0,1} ref lists. */ + init_reflist_builder(ctx, &reflist_builder); + build_p_ref_list(&reflist_builder, h264_ctx->reflists.p); + build_b_ref_lists(&reflist_builder, h264_ctx->reflists.b0, + h264_ctx->reflists.b1); + return 0; +} + +void hantro_h264_dec_exit(struct hantro_ctx *ctx) +{ + struct hantro_dev *vpu = ctx->dev; + struct hantro_h264_dec_hw_ctx *h264_dec = &ctx->h264_dec; + struct hantro_aux_buf *priv = &h264_dec->priv; + + dma_free_coherent(vpu->dev, priv->size, priv->cpu, priv->dma); +} + +int hantro_h264_dec_init(struct hantro_ctx *ctx) +{ + struct hantro_dev *vpu = ctx->dev; + struct hantro_h264_dec_hw_ctx *h264_dec = &ctx->h264_dec; + struct hantro_aux_buf *priv = &h264_dec->priv; + struct hantro_h264_dec_priv_tbl *tbl; + struct v4l2_pix_format_mplane pix_mp; + + priv->cpu = dma_alloc_coherent(vpu->dev, sizeof(*tbl), &priv->dma, + GFP_KERNEL); + if (!priv->cpu) + return -ENOMEM; + + priv->size = sizeof(*tbl); + tbl = priv->cpu; + memcpy(tbl->cabac_table, h264_cabac_table, sizeof(tbl->cabac_table)); + + v4l2_fill_pixfmt_mp(&pix_mp, ctx->dst_fmt.pixelformat, + ctx->dst_fmt.width, ctx->dst_fmt.height); + h264_dec->pic_size = pix_mp.plane_fmt[0].sizeimage; + + return 0; +} diff --git a/drivers/staging/media/hantro/hantro_hw.h b/drivers/staging/media/hantro/hantro_hw.h index 3c361c2e9b88..2fab655bf098 100644 --- a/drivers/staging/media/hantro/hantro_hw.h +++ b/drivers/staging/media/hantro/hantro_hw.h @@ -11,9 +11,13 @@ #include <linux/interrupt.h> #include <linux/v4l2-controls.h> +#include <media/h264-ctrls.h> #include <media/mpeg2-ctrls.h> +#include <media/vp8-ctrls.h> #include <media/videobuf2-core.h> +#define DEC_8190_ALIGN_MASK 0x07U + struct hantro_dev; struct hantro_ctx; struct hantro_buf; @@ -39,6 +43,54 @@ struct hantro_jpeg_enc_hw_ctx { struct hantro_aux_buf bounce_buffer; }; +/* Max. number of entries in the DPB (HW limitation). */ +#define HANTRO_H264_DPB_SIZE 16 + +/** + * struct hantro_h264_dec_ctrls + * @decode: Decode params + * @scaling: Scaling info + * @slice: Slice params + * @sps: SPS info + * @pps: PPS info + */ +struct hantro_h264_dec_ctrls { + const struct v4l2_ctrl_h264_decode_params *decode; + const struct v4l2_ctrl_h264_scaling_matrix *scaling; + const struct v4l2_ctrl_h264_slice_params *slices; + const struct v4l2_ctrl_h264_sps *sps; + const struct v4l2_ctrl_h264_pps *pps; +}; + +/** + * struct hantro_h264_dec_reflists + * @p: P reflist + * @b0: B0 reflist + * @b1: B1 reflist + */ +struct hantro_h264_dec_reflists { + u8 p[HANTRO_H264_DPB_SIZE]; + u8 b0[HANTRO_H264_DPB_SIZE]; + u8 b1[HANTRO_H264_DPB_SIZE]; +}; + +/** + * struct hantro_h264_dec_hw_ctx + * @priv: Private auxiliary buffer for hardware. + * @dpb: DPB + * @reflists: P/B0/B1 reflists + * @ctrls: V4L2 controls attached to a run + * @pic_size: Size in bytes of decoded picture, this is needed + * to pass the location of motion vectors. + */ +struct hantro_h264_dec_hw_ctx { + struct hantro_aux_buf priv; + struct v4l2_h264_dpb_entry dpb[HANTRO_H264_DPB_SIZE]; + struct hantro_h264_dec_reflists reflists; + struct hantro_h264_dec_ctrls ctrls; + size_t pic_size; +}; + /** * struct hantro_mpeg2_dec_hw_ctx * @qtable: Quantization table @@ -48,6 +100,16 @@ struct hantro_mpeg2_dec_hw_ctx { }; /** + * struct hantro_vp8d_hw_ctx + * @segment_map: Segment map buffer. + * @prob_tbl: Probability table buffer. + */ +struct hantro_vp8_dec_hw_ctx { + struct hantro_aux_buf segment_map; + struct hantro_aux_buf prob_tbl; +}; + +/** * struct hantro_codec_ops - codec mode specific operations * * @init: If needed, can be used for initialization. @@ -82,16 +144,27 @@ extern const struct hantro_variant rk3399_vpu_variant; extern const struct hantro_variant rk3328_vpu_variant; extern const struct hantro_variant rk3288_vpu_variant; +extern const u32 hantro_vp8_dec_mc_filter[8][6]; + void hantro_watchdog(struct work_struct *work); void hantro_run(struct hantro_ctx *ctx); void hantro_irq_done(struct hantro_dev *vpu, unsigned int bytesused, enum vb2_buffer_state result); +void hantro_prepare_run(struct hantro_ctx *ctx); +void hantro_finish_run(struct hantro_ctx *ctx); void hantro_h1_jpeg_enc_run(struct hantro_ctx *ctx); void rk3399_vpu_jpeg_enc_run(struct hantro_ctx *ctx); int hantro_jpeg_enc_init(struct hantro_ctx *ctx); void hantro_jpeg_enc_exit(struct hantro_ctx *ctx); +struct vb2_buffer *hantro_h264_get_ref_buf(struct hantro_ctx *ctx, + unsigned int dpb_idx); +int hantro_h264_dec_prepare_run(struct hantro_ctx *ctx); +void hantro_g1_h264_dec_run(struct hantro_ctx *ctx); +int hantro_h264_dec_init(struct hantro_ctx *ctx); +void hantro_h264_dec_exit(struct hantro_ctx *ctx); + void hantro_g1_mpeg2_dec_run(struct hantro_ctx *ctx); void rk3399_vpu_mpeg2_dec_run(struct hantro_ctx *ctx); void hantro_mpeg2_dec_copy_qtable(u8 *qtable, @@ -99,4 +172,11 @@ void hantro_mpeg2_dec_copy_qtable(u8 *qtable, int hantro_mpeg2_dec_init(struct hantro_ctx *ctx); void hantro_mpeg2_dec_exit(struct hantro_ctx *ctx); +void hantro_g1_vp8_dec_run(struct hantro_ctx *ctx); +void rk3399_vpu_vp8_dec_run(struct hantro_ctx *ctx); +int hantro_vp8_dec_init(struct hantro_ctx *ctx); +void hantro_vp8_dec_exit(struct hantro_ctx *ctx); +void hantro_vp8_prob_update(struct hantro_ctx *ctx, + const struct v4l2_ctrl_vp8_frame_header *hdr); + #endif /* HANTRO_HW_H_ */ diff --git a/drivers/staging/media/hantro/hantro_v4l2.c b/drivers/staging/media/hantro/hantro_v4l2.c index 68f45ee66821..3dae52abb96c 100644 --- a/drivers/staging/media/hantro/hantro_v4l2.c +++ b/drivers/staging/media/hantro/hantro_v4l2.c @@ -239,6 +239,15 @@ static int vidioc_try_fmt(struct file *file, void *priv, struct v4l2_format *f, /* Fill remaining fields */ v4l2_fill_pixfmt_mp(pix_mp, fmt->fourcc, pix_mp->width, pix_mp->height); + /* + * The H264 decoder needs extra space on the output buffers + * to store motion vectors. This is needed for reference + * frames. + */ + if (ctx->vpu_src_fmt->fourcc == V4L2_PIX_FMT_H264_SLICE) + pix_mp->plane_fmt[0].sizeimage += + 128 * DIV_ROUND_UP(pix_mp->width, 16) * + DIV_ROUND_UP(pix_mp->height, 16); } else if (!pix_mp->plane_fmt[0].sizeimage) { /* * For coded formats the application can specify @@ -344,6 +353,8 @@ hantro_update_requires_request(struct hantro_ctx *ctx, u32 fourcc) ctx->fh.m2m_ctx->out_q_ctx.q.requires_requests = false; break; case V4L2_PIX_FMT_MPEG2_SLICE: + case V4L2_PIX_FMT_VP8_FRAME: + case V4L2_PIX_FMT_H264_SLICE: ctx->fh.m2m_ctx->out_q_ctx.q.requires_requests = true; break; default: diff --git a/drivers/staging/media/hantro/hantro_vp8.c b/drivers/staging/media/hantro/hantro_vp8.c new file mode 100644 index 000000000000..0e02d147b189 --- /dev/null +++ b/drivers/staging/media/hantro/hantro_vp8.c @@ -0,0 +1,201 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Hantro VPU codec driver + * + * Copyright (C) 2018 Rockchip Electronics Co., Ltd. + */ + +#include "hantro.h" + +/* + * probs table with packed + */ +struct vp8_prob_tbl_packed { + u8 prob_mb_skip_false; + u8 prob_intra; + u8 prob_ref_last; + u8 prob_ref_golden; + u8 prob_segment[3]; + u8 padding0; + + u8 prob_luma_16x16_pred_mode[4]; + u8 prob_chroma_pred_mode[3]; + u8 padding1; + + /* mv prob */ + u8 prob_mv_context[2][19]; + u8 padding2[2]; + + /* coeff probs */ + u8 prob_coeffs[4][8][3][11]; + u8 padding3[96]; +}; + +/* + * filter taps taken to 7-bit precision, + * reference RFC6386#Page-16, filters[8][6] + */ +const u32 hantro_vp8_dec_mc_filter[8][6] = { + { 0, 0, 128, 0, 0, 0 }, + { 0, -6, 123, 12, -1, 0 }, + { 2, -11, 108, 36, -8, 1 }, + { 0, -9, 93, 50, -6, 0 }, + { 3, -16, 77, 77, -16, 3 }, + { 0, -6, 50, 93, -9, 0 }, + { 1, -8, 36, 108, -11, 2 }, + { 0, -1, 12, 123, -6, 0 } +}; + +void hantro_vp8_prob_update(struct hantro_ctx *ctx, + const struct v4l2_ctrl_vp8_frame_header *hdr) +{ + const struct v4l2_vp8_entropy_header *entropy = &hdr->entropy_header; + u32 i, j, k; + u8 *dst; + + /* first probs */ + dst = ctx->vp8_dec.prob_tbl.cpu; + + dst[0] = hdr->prob_skip_false; + dst[1] = hdr->prob_intra; + dst[2] = hdr->prob_last; + dst[3] = hdr->prob_gf; + dst[4] = hdr->segment_header.segment_probs[0]; + dst[5] = hdr->segment_header.segment_probs[1]; + dst[6] = hdr->segment_header.segment_probs[2]; + dst[7] = 0; + + dst += 8; + dst[0] = entropy->y_mode_probs[0]; + dst[1] = entropy->y_mode_probs[1]; + dst[2] = entropy->y_mode_probs[2]; + dst[3] = entropy->y_mode_probs[3]; + dst[4] = entropy->uv_mode_probs[0]; + dst[5] = entropy->uv_mode_probs[1]; + dst[6] = entropy->uv_mode_probs[2]; + dst[7] = 0; /*unused */ + + /* mv probs */ + dst += 8; + dst[0] = entropy->mv_probs[0][0]; /* is short */ + dst[1] = entropy->mv_probs[1][0]; + dst[2] = entropy->mv_probs[0][1]; /* sign */ + dst[3] = entropy->mv_probs[1][1]; + dst[4] = entropy->mv_probs[0][8 + 9]; + dst[5] = entropy->mv_probs[0][9 + 9]; + dst[6] = entropy->mv_probs[1][8 + 9]; + dst[7] = entropy->mv_probs[1][9 + 9]; + dst += 8; + for (i = 0; i < 2; ++i) { + for (j = 0; j < 8; j += 4) { + dst[0] = entropy->mv_probs[i][j + 9 + 0]; + dst[1] = entropy->mv_probs[i][j + 9 + 1]; + dst[2] = entropy->mv_probs[i][j + 9 + 2]; + dst[3] = entropy->mv_probs[i][j + 9 + 3]; + dst += 4; + } + } + for (i = 0; i < 2; ++i) { + dst[0] = entropy->mv_probs[i][0 + 2]; + dst[1] = entropy->mv_probs[i][1 + 2]; + dst[2] = entropy->mv_probs[i][2 + 2]; + dst[3] = entropy->mv_probs[i][3 + 2]; + dst[4] = entropy->mv_probs[i][4 + 2]; + dst[5] = entropy->mv_probs[i][5 + 2]; + dst[6] = entropy->mv_probs[i][6 + 2]; + dst[7] = 0; /*unused */ + dst += 8; + } + + /* coeff probs (header part) */ + dst = ctx->vp8_dec.prob_tbl.cpu; + dst += (8 * 7); + for (i = 0; i < 4; ++i) { + for (j = 0; j < 8; ++j) { + for (k = 0; k < 3; ++k) { + dst[0] = entropy->coeff_probs[i][j][k][0]; + dst[1] = entropy->coeff_probs[i][j][k][1]; + dst[2] = entropy->coeff_probs[i][j][k][2]; + dst[3] = entropy->coeff_probs[i][j][k][3]; + dst += 4; + } + } + } + + /* coeff probs (footer part) */ + dst = ctx->vp8_dec.prob_tbl.cpu; + dst += (8 * 55); + for (i = 0; i < 4; ++i) { + for (j = 0; j < 8; ++j) { + for (k = 0; k < 3; ++k) { + dst[0] = entropy->coeff_probs[i][j][k][4]; + dst[1] = entropy->coeff_probs[i][j][k][5]; + dst[2] = entropy->coeff_probs[i][j][k][6]; + dst[3] = entropy->coeff_probs[i][j][k][7]; + dst[4] = entropy->coeff_probs[i][j][k][8]; + dst[5] = entropy->coeff_probs[i][j][k][9]; + dst[6] = entropy->coeff_probs[i][j][k][10]; + dst[7] = 0; /*unused */ + dst += 8; + } + } + } +} + +int hantro_vp8_dec_init(struct hantro_ctx *ctx) +{ + struct hantro_dev *vpu = ctx->dev; + struct hantro_aux_buf *aux_buf; + unsigned int mb_width, mb_height; + size_t segment_map_size; + int ret; + + /* segment map table size calculation */ + mb_width = DIV_ROUND_UP(ctx->dst_fmt.width, 16); + mb_height = DIV_ROUND_UP(ctx->dst_fmt.height, 16); + segment_map_size = round_up(DIV_ROUND_UP(mb_width * mb_height, 4), 64); + + /* + * In context init the dma buffer for segment map must be allocated. + * And the data in segment map buffer must be set to all zero. + */ + aux_buf = &ctx->vp8_dec.segment_map; + aux_buf->size = segment_map_size; + aux_buf->cpu = dma_alloc_coherent(vpu->dev, aux_buf->size, + &aux_buf->dma, GFP_KERNEL); + if (!aux_buf->cpu) + return -ENOMEM; + + /* + * Allocate probability table buffer, + * total 1208 bytes, 4K page is far enough. + */ + aux_buf = &ctx->vp8_dec.prob_tbl; + aux_buf->size = sizeof(struct vp8_prob_tbl_packed); + aux_buf->cpu = dma_alloc_coherent(vpu->dev, aux_buf->size, + &aux_buf->dma, GFP_KERNEL); + if (!aux_buf->cpu) { + ret = -ENOMEM; + goto err_free_seg_map; + } + + return 0; + +err_free_seg_map: + dma_free_coherent(vpu->dev, ctx->vp8_dec.segment_map.size, + ctx->vp8_dec.segment_map.cpu, + ctx->vp8_dec.segment_map.dma); + + return ret; +} + +void hantro_vp8_dec_exit(struct hantro_ctx *ctx) +{ + struct hantro_vp8_dec_hw_ctx *vp8_dec = &ctx->vp8_dec; + struct hantro_dev *vpu = ctx->dev; + + dma_free_coherent(vpu->dev, vp8_dec->segment_map.size, + vp8_dec->segment_map.cpu, vp8_dec->segment_map.dma); + dma_free_coherent(vpu->dev, vp8_dec->prob_tbl.size, + vp8_dec->prob_tbl.cpu, vp8_dec->prob_tbl.dma); +} diff --git a/drivers/staging/media/hantro/rk3288_vpu_hw.c b/drivers/staging/media/hantro/rk3288_vpu_hw.c index bcacc4f51093..6bfcc47d1e58 100644 --- a/drivers/staging/media/hantro/rk3288_vpu_hw.c +++ b/drivers/staging/media/hantro/rk3288_vpu_hw.c @@ -62,6 +62,19 @@ static const struct hantro_fmt rk3288_vpu_dec_fmts[] = { .codec_mode = HANTRO_MODE_NONE, }, { + .fourcc = V4L2_PIX_FMT_H264_SLICE, + .codec_mode = HANTRO_MODE_H264_DEC, + .max_depth = 2, + .frmsize = { + .min_width = 48, + .max_width = 3840, + .step_width = H264_MB_DIM, + .min_height = 48, + .max_height = 2160, + .step_height = H264_MB_DIM, + }, + }, + { .fourcc = V4L2_PIX_FMT_MPEG2_SLICE, .codec_mode = HANTRO_MODE_MPEG2_DEC, .max_depth = 2, @@ -74,6 +87,19 @@ static const struct hantro_fmt rk3288_vpu_dec_fmts[] = { .step_height = MPEG2_MB_DIM, }, }, + { + .fourcc = V4L2_PIX_FMT_VP8_FRAME, + .codec_mode = HANTRO_MODE_VP8_DEC, + .max_depth = 2, + .frmsize = { + .min_width = 48, + .max_width = 3840, + .step_width = VP8_MB_DIM, + .min_height = 48, + .max_height = 2160, + .step_height = VP8_MB_DIM, + }, + }, }; static irqreturn_t rk3288_vepu_irq(int irq, void *dev_id) @@ -149,12 +175,24 @@ static const struct hantro_codec_ops rk3288_vpu_codec_ops[] = { .init = hantro_jpeg_enc_init, .exit = hantro_jpeg_enc_exit, }, + [HANTRO_MODE_H264_DEC] = { + .run = hantro_g1_h264_dec_run, + .reset = rk3288_vpu_dec_reset, + .init = hantro_h264_dec_init, + .exit = hantro_h264_dec_exit, + }, [HANTRO_MODE_MPEG2_DEC] = { .run = hantro_g1_mpeg2_dec_run, .reset = rk3288_vpu_dec_reset, .init = hantro_mpeg2_dec_init, .exit = hantro_mpeg2_dec_exit, }, + [HANTRO_MODE_VP8_DEC] = { + .run = hantro_g1_vp8_dec_run, + .reset = rk3288_vpu_dec_reset, + .init = hantro_vp8_dec_init, + .exit = hantro_vp8_dec_exit, + }, }; /* @@ -177,7 +215,8 @@ const struct hantro_variant rk3288_vpu_variant = { .dec_offset = 0x400, .dec_fmts = rk3288_vpu_dec_fmts, .num_dec_fmts = ARRAY_SIZE(rk3288_vpu_dec_fmts), - .codec = HANTRO_JPEG_ENCODER | HANTRO_MPEG2_DECODER, + .codec = HANTRO_JPEG_ENCODER | HANTRO_MPEG2_DECODER | + HANTRO_VP8_DECODER | HANTRO_H264_DECODER, .codec_ops = rk3288_vpu_codec_ops, .irqs = rk3288_irqs, .num_irqs = ARRAY_SIZE(rk3288_irqs), diff --git a/drivers/staging/media/hantro/rk3399_vpu_hw.c b/drivers/staging/media/hantro/rk3399_vpu_hw.c index 5718f8063542..14d14bc6b12b 100644 --- a/drivers/staging/media/hantro/rk3399_vpu_hw.c +++ b/drivers/staging/media/hantro/rk3399_vpu_hw.c @@ -73,6 +73,19 @@ static const struct hantro_fmt rk3399_vpu_dec_fmts[] = { .step_height = MPEG2_MB_DIM, }, }, + { + .fourcc = V4L2_PIX_FMT_VP8_FRAME, + .codec_mode = HANTRO_MODE_VP8_DEC, + .max_depth = 2, + .frmsize = { + .min_width = 48, + .max_width = 3840, + .step_width = VP8_MB_DIM, + .min_height = 48, + .max_height = 2160, + .step_height = VP8_MB_DIM, + }, + }, }; static irqreturn_t rk3399_vepu_irq(int irq, void *dev_id) @@ -154,6 +167,12 @@ static const struct hantro_codec_ops rk3399_vpu_codec_ops[] = { .init = hantro_mpeg2_dec_init, .exit = hantro_mpeg2_dec_exit, }, + [HANTRO_MODE_VP8_DEC] = { + .run = rk3399_vpu_vp8_dec_run, + .reset = rk3399_vpu_dec_reset, + .init = hantro_vp8_dec_init, + .exit = hantro_vp8_dec_exit, + }, }; /* @@ -176,7 +195,8 @@ const struct hantro_variant rk3399_vpu_variant = { .dec_offset = 0x400, .dec_fmts = rk3399_vpu_dec_fmts, .num_dec_fmts = ARRAY_SIZE(rk3399_vpu_dec_fmts), - .codec = HANTRO_JPEG_ENCODER | HANTRO_MPEG2_DECODER, + .codec = HANTRO_JPEG_ENCODER | HANTRO_MPEG2_DECODER | + HANTRO_VP8_DECODER, .codec_ops = rk3399_vpu_codec_ops, .irqs = rk3399_irqs, .num_irqs = ARRAY_SIZE(rk3399_irqs), @@ -184,3 +204,20 @@ const struct hantro_variant rk3399_vpu_variant = { .clk_names = rk3399_clk_names, .num_clocks = ARRAY_SIZE(rk3399_clk_names) }; + +static const struct hantro_irq rk3328_irqs[] = { + { "vdpu", rk3399_vdpu_irq }, +}; + +const struct hantro_variant rk3328_vpu_variant = { + .dec_offset = 0x400, + .dec_fmts = rk3399_vpu_dec_fmts, + .num_dec_fmts = ARRAY_SIZE(rk3399_vpu_dec_fmts), + .codec = HANTRO_MPEG2_DECODER | HANTRO_VP8_DECODER, + .codec_ops = rk3399_vpu_codec_ops, + .irqs = rk3328_irqs, + .num_irqs = ARRAY_SIZE(rk3328_irqs), + .init = rk3399_vpu_hw_init, + .clk_names = rk3399_clk_names, + .num_clocks = ARRAY_SIZE(rk3399_clk_names), +}; diff --git a/drivers/staging/media/hantro/rk3399_vpu_hw_jpeg_enc.c b/drivers/staging/media/hantro/rk3399_vpu_hw_jpeg_enc.c index ae66354d2d93..06162f569b5e 100644 --- a/drivers/staging/media/hantro/rk3399_vpu_hw_jpeg_enc.c +++ b/drivers/staging/media/hantro/rk3399_vpu_hw_jpeg_enc.c @@ -113,14 +113,12 @@ void rk3399_vpu_jpeg_enc_run(struct hantro_ctx *ctx) struct hantro_dev *vpu = ctx->dev; struct vb2_v4l2_buffer *src_buf, *dst_buf; struct hantro_jpeg_ctx jpeg_ctx; - struct media_request *src_req; u32 reg; - src_buf = v4l2_m2m_next_src_buf(ctx->fh.m2m_ctx); - dst_buf = v4l2_m2m_next_dst_buf(ctx->fh.m2m_ctx); + src_buf = hantro_get_src_buf(ctx); + dst_buf = hantro_get_dst_buf(ctx); - src_req = src_buf->vb2_buf.req_obj.req; - v4l2_ctrl_request_setup(src_req, &ctx->ctrl_handler); + hantro_prepare_run(ctx); memset(&jpeg_ctx, 0, sizeof(jpeg_ctx)); jpeg_ctx.buffer = vb2_plane_vaddr(&dst_buf->vb2_buf, 0); @@ -157,9 +155,7 @@ void rk3399_vpu_jpeg_enc_run(struct hantro_ctx *ctx) | VEPU_REG_ENCODE_FORMAT_JPEG | VEPU_REG_ENCODE_ENABLE; - v4l2_ctrl_request_complete(src_req, &ctx->ctrl_handler); - /* Kick the watchdog and start encoding */ - schedule_delayed_work(&vpu->watchdog_work, msecs_to_jiffies(2000)); + hantro_finish_run(ctx); vepu_write(vpu, reg, VEPU_REG_ENCODE_START); } diff --git a/drivers/staging/media/hantro/rk3399_vpu_hw_mpeg2_dec.c b/drivers/staging/media/hantro/rk3399_vpu_hw_mpeg2_dec.c index 8685bddfbcab..e7ba5c0441cc 100644 --- a/drivers/staging/media/hantro/rk3399_vpu_hw_mpeg2_dec.c +++ b/drivers/staging/media/hantro/rk3399_vpu_hw_mpeg2_dec.c @@ -169,12 +169,10 @@ void rk3399_vpu_mpeg2_dec_run(struct hantro_ctx *ctx) const struct v4l2_mpeg2_picture *picture; u32 reg; - src_buf = v4l2_m2m_next_src_buf(ctx->fh.m2m_ctx); - dst_buf = v4l2_m2m_next_dst_buf(ctx->fh.m2m_ctx); + src_buf = hantro_get_src_buf(ctx); + dst_buf = hantro_get_dst_buf(ctx); - /* Apply request controls if any */ - v4l2_ctrl_request_setup(src_buf->vb2_buf.req_obj.req, - &ctx->ctrl_handler); + hantro_prepare_run(ctx); slice_params = hantro_get_ctrl(ctx, V4L2_CID_MPEG_VIDEO_MPEG2_SLICE_PARAMS); @@ -254,12 +252,8 @@ void rk3399_vpu_mpeg2_dec_run(struct hantro_ctx *ctx) &dst_buf->vb2_buf, sequence, picture, slice_params); - /* Controls no longer in-use, we can complete them */ - v4l2_ctrl_request_complete(src_buf->vb2_buf.req_obj.req, - &ctx->ctrl_handler); - /* Kick the watchdog and start decoding */ - schedule_delayed_work(&vpu->watchdog_work, msecs_to_jiffies(2000)); + hantro_finish_run(ctx); reg = vdpu_read(vpu, VDPU_SWREG(57)) | VDPU_REG_DEC_E(1); vdpu_write(vpu, reg, VDPU_SWREG(57)); diff --git a/drivers/staging/media/hantro/rk3399_vpu_hw_vp8_dec.c b/drivers/staging/media/hantro/rk3399_vpu_hw_vp8_dec.c new file mode 100644 index 000000000000..f17e32620b08 --- /dev/null +++ b/drivers/staging/media/hantro/rk3399_vpu_hw_vp8_dec.c @@ -0,0 +1,595 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Rockchip VPU codec vp8 decode driver + * + * Copyright (C) 2014 Rockchip Electronics Co., Ltd. + * ZhiChao Yu <zhichao.yu@rock-chips.com> + * + * Copyright (C) 2014 Google LLC. + * Tomasz Figa <tfiga@chromium.org> + * + * Copyright (C) 2015 Rockchip Electronics Co., Ltd. + * Alpha Lin <alpha.lin@rock-chips.com> + */ + +#include <media/v4l2-mem2mem.h> +#include <media/vp8-ctrls.h> + +#include "hantro_hw.h" +#include "hantro.h" +#include "hantro_g1_regs.h" + +#define VDPU_REG_DEC_CTRL0 0x0c8 +#define VDPU_REG_STREAM_LEN 0x0cc +#define VDPU_REG_DEC_FORMAT 0x0d4 +#define VDPU_REG_DEC_CTRL0_DEC_MODE(x) (((x) & 0xf) << 0) +#define VDPU_REG_DATA_ENDIAN 0x0d8 +#define VDPU_REG_CONFIG_DEC_STRENDIAN_E BIT(5) +#define VDPU_REG_CONFIG_DEC_STRSWAP32_E BIT(4) +#define VDPU_REG_CONFIG_DEC_OUTSWAP32_E BIT(3) +#define VDPU_REG_CONFIG_DEC_INSWAP32_E BIT(2) +#define VDPU_REG_CONFIG_DEC_OUT_ENDIAN BIT(1) +#define VDPU_REG_CONFIG_DEC_IN_ENDIAN BIT(0) +#define VDPU_REG_AXI_CTRL 0x0e0 +#define VDPU_REG_CONFIG_DEC_MAX_BURST(x) (((x) & 0x1f) << 16) +#define VDPU_REG_EN_FLAGS 0x0e4 +#define VDPU_REG_DEC_CTRL0_PIC_INTER_E BIT(14) +#define VDPU_REG_CONFIG_DEC_TIMEOUT_E BIT(5) +#define VDPU_REG_CONFIG_DEC_CLK_GATE_E BIT(4) +#define VDPU_REG_PRED_FLT 0x0ec +#define VDPU_REG_ADDR_QTABLE 0x0f4 +#define VDPU_REG_ADDR_DST 0x0fc +#define VDPU_REG_ADDR_STR 0x100 +#define VDPU_REG_VP8_PIC_MB_SIZE 0x1e0 +#define VDPU_REG_VP8_DCT_START_BIT 0x1e4 +#define VDPU_REG_DEC_CTRL4_VC1_HEIGHT_EXT BIT(13) +#define VDPU_REG_DEC_CTRL4_BILIN_MC_E BIT(12) +#define VDPU_REG_VP8_CTRL0 0x1e8 +#define VDPU_REG_VP8_DATA_VAL 0x1f0 +#define VDPU_REG_PRED_FLT7 0x1f4 +#define VDPU_REG_PRED_FLT8 0x1f8 +#define VDPU_REG_PRED_FLT9 0x1fc +#define VDPU_REG_PRED_FLT10 0x200 +#define VDPU_REG_FILTER_LEVEL 0x204 +#define VDPU_REG_VP8_QUANTER0 0x208 +#define VDPU_REG_VP8_ADDR_REF0 0x20c +#define VDPU_REG_FILTER_MB_ADJ 0x210 +#define VDPU_REG_REF_PIC_FILT_TYPE_E BIT(31) +#define VDPU_REG_REF_PIC_FILT_SHARPNESS(x) (((x) & 0x7) << 28) +#define VDPU_REG_FILTER_REF_ADJ 0x214 +#define VDPU_REG_VP8_ADDR_REF2_5(i) (0x218 + ((i) * 0x4)) +#define VDPU_REG_VP8_GREF_SIGN_BIAS BIT(0) +#define VDPU_REG_VP8_AREF_SIGN_BIAS BIT(0) +#define VDPU_REG_VP8_DCT_BASE(i) \ + (0x230 + ((((i) < 5) ? (i) : ((i) + 1)) * 0x4)) +#define VDPU_REG_VP8_ADDR_CTRL_PART 0x244 +#define VDPU_REG_VP8_SEGMENT_VAL 0x254 +#define VDPU_REG_FWD_PIC1_SEGMENT_BASE(x) ((x) << 0) +#define VDPU_REG_FWD_PIC1_SEGMENT_UPD_E BIT(1) +#define VDPU_REG_FWD_PIC1_SEGMENT_E BIT(0) +#define VDPU_REG_VP8_DCT_START_BIT2 0x258 +#define VDPU_REG_VP8_QUANTER1 0x25c +#define VDPU_REG_VP8_QUANTER2 0x260 +#define VDPU_REG_PRED_FLT1 0x264 +#define VDPU_REG_PRED_FLT2 0x268 +#define VDPU_REG_PRED_FLT3 0x26c +#define VDPU_REG_PRED_FLT4 0x270 +#define VDPU_REG_PRED_FLT5 0x274 +#define VDPU_REG_PRED_FLT6 0x278 + +static const struct hantro_reg vp8_dec_dct_base[8] = { + { VDPU_REG_ADDR_STR, 0, 0xffffffff }, + { VDPU_REG_VP8_DCT_BASE(0), 0, 0xffffffff }, + { VDPU_REG_VP8_DCT_BASE(1), 0, 0xffffffff }, + { VDPU_REG_VP8_DCT_BASE(2), 0, 0xffffffff }, + { VDPU_REG_VP8_DCT_BASE(3), 0, 0xffffffff }, + { VDPU_REG_VP8_DCT_BASE(4), 0, 0xffffffff }, + { VDPU_REG_VP8_DCT_BASE(5), 0, 0xffffffff }, + { VDPU_REG_VP8_DCT_BASE(6), 0, 0xffffffff }, +}; + +static const struct hantro_reg vp8_dec_lf_level[4] = { + { VDPU_REG_FILTER_LEVEL, 18, 0x3f }, + { VDPU_REG_FILTER_LEVEL, 12, 0x3f }, + { VDPU_REG_FILTER_LEVEL, 6, 0x3f }, + { VDPU_REG_FILTER_LEVEL, 0, 0x3f }, +}; + +static const struct hantro_reg vp8_dec_mb_adj[4] = { + { VDPU_REG_FILTER_MB_ADJ, 21, 0x7f }, + { VDPU_REG_FILTER_MB_ADJ, 14, 0x7f }, + { VDPU_REG_FILTER_MB_ADJ, 7, 0x7f }, + { VDPU_REG_FILTER_MB_ADJ, 0, 0x7f }, +}; + +static const struct hantro_reg vp8_dec_ref_adj[4] = { + { VDPU_REG_FILTER_REF_ADJ, 21, 0x7f }, + { VDPU_REG_FILTER_REF_ADJ, 14, 0x7f }, + { VDPU_REG_FILTER_REF_ADJ, 7, 0x7f }, + { VDPU_REG_FILTER_REF_ADJ, 0, 0x7f }, +}; + +static const struct hantro_reg vp8_dec_quant[4] = { + { VDPU_REG_VP8_QUANTER0, 11, 0x7ff }, + { VDPU_REG_VP8_QUANTER0, 0, 0x7ff }, + { VDPU_REG_VP8_QUANTER1, 11, 0x7ff }, + { VDPU_REG_VP8_QUANTER1, 0, 0x7ff }, +}; + +static const struct hantro_reg vp8_dec_quant_delta[5] = { + { VDPU_REG_VP8_QUANTER0, 27, 0x1f }, + { VDPU_REG_VP8_QUANTER0, 22, 0x1f }, + { VDPU_REG_VP8_QUANTER1, 27, 0x1f }, + { VDPU_REG_VP8_QUANTER1, 22, 0x1f }, + { VDPU_REG_VP8_QUANTER2, 27, 0x1f }, +}; + +static const struct hantro_reg vp8_dec_dct_start_bits[8] = { + { VDPU_REG_VP8_CTRL0, 26, 0x3f }, + { VDPU_REG_VP8_DCT_START_BIT, 26, 0x3f }, + { VDPU_REG_VP8_DCT_START_BIT, 20, 0x3f }, + { VDPU_REG_VP8_DCT_START_BIT2, 24, 0x3f }, + { VDPU_REG_VP8_DCT_START_BIT2, 18, 0x3f }, + { VDPU_REG_VP8_DCT_START_BIT2, 12, 0x3f }, + { VDPU_REG_VP8_DCT_START_BIT2, 6, 0x3f }, + { VDPU_REG_VP8_DCT_START_BIT2, 0, 0x3f }, +}; + +static const struct hantro_reg vp8_dec_pred_bc_tap[8][6] = { + { + { 0, 0, 0}, + { VDPU_REG_PRED_FLT, 22, 0x3ff }, + { VDPU_REG_PRED_FLT, 12, 0x3ff }, + { VDPU_REG_PRED_FLT, 2, 0x3ff }, + { VDPU_REG_PRED_FLT1, 22, 0x3ff }, + { 0, 0, 0}, + }, { + { 0, 0, 0}, + { VDPU_REG_PRED_FLT1, 12, 0x3ff }, + { VDPU_REG_PRED_FLT1, 2, 0x3ff }, + { VDPU_REG_PRED_FLT2, 22, 0x3ff }, + { VDPU_REG_PRED_FLT2, 12, 0x3ff }, + { 0, 0, 0}, + }, { + { VDPU_REG_PRED_FLT10, 10, 0x3 }, + { VDPU_REG_PRED_FLT2, 2, 0x3ff }, + { VDPU_REG_PRED_FLT3, 22, 0x3ff }, + { VDPU_REG_PRED_FLT3, 12, 0x3ff }, + { VDPU_REG_PRED_FLT3, 2, 0x3ff }, + { VDPU_REG_PRED_FLT10, 8, 0x3}, + }, { + { 0, 0, 0}, + { VDPU_REG_PRED_FLT4, 22, 0x3ff }, + { VDPU_REG_PRED_FLT4, 12, 0x3ff }, + { VDPU_REG_PRED_FLT4, 2, 0x3ff }, + { VDPU_REG_PRED_FLT5, 22, 0x3ff }, + { 0, 0, 0}, + }, { + { VDPU_REG_PRED_FLT10, 6, 0x3 }, + { VDPU_REG_PRED_FLT5, 12, 0x3ff }, + { VDPU_REG_PRED_FLT5, 2, 0x3ff }, + { VDPU_REG_PRED_FLT6, 22, 0x3ff }, + { VDPU_REG_PRED_FLT6, 12, 0x3ff }, + { VDPU_REG_PRED_FLT10, 4, 0x3 }, + }, { + { 0, 0, 0}, + { VDPU_REG_PRED_FLT6, 2, 0x3ff }, + { VDPU_REG_PRED_FLT7, 22, 0x3ff }, + { VDPU_REG_PRED_FLT7, 12, 0x3ff }, + { VDPU_REG_PRED_FLT7, 2, 0x3ff }, + { 0, 0, 0}, + }, { + { VDPU_REG_PRED_FLT10, 2, 0x3 }, + { VDPU_REG_PRED_FLT8, 22, 0x3ff }, + { VDPU_REG_PRED_FLT8, 12, 0x3ff }, + { VDPU_REG_PRED_FLT8, 2, 0x3ff }, + { VDPU_REG_PRED_FLT9, 22, 0x3ff }, + { VDPU_REG_PRED_FLT10, 0, 0x3 }, + }, { + { 0, 0, 0}, + { VDPU_REG_PRED_FLT9, 12, 0x3ff }, + { VDPU_REG_PRED_FLT9, 2, 0x3ff }, + { VDPU_REG_PRED_FLT10, 22, 0x3ff }, + { VDPU_REG_PRED_FLT10, 12, 0x3ff }, + { 0, 0, 0}, + }, +}; + +static const struct hantro_reg vp8_dec_mb_start_bit = { + .base = VDPU_REG_VP8_CTRL0, + .shift = 18, + .mask = 0x3f +}; + +static const struct hantro_reg vp8_dec_mb_aligned_data_len = { + .base = VDPU_REG_VP8_DATA_VAL, + .shift = 0, + .mask = 0x3fffff +}; + +static const struct hantro_reg vp8_dec_num_dct_partitions = { + .base = VDPU_REG_VP8_DATA_VAL, + .shift = 24, + .mask = 0xf +}; + +static const struct hantro_reg vp8_dec_stream_len = { + .base = VDPU_REG_STREAM_LEN, + .shift = 0, + .mask = 0xffffff +}; + +static const struct hantro_reg vp8_dec_mb_width = { + .base = VDPU_REG_VP8_PIC_MB_SIZE, + .shift = 23, + .mask = 0x1ff +}; + +static const struct hantro_reg vp8_dec_mb_height = { + .base = VDPU_REG_VP8_PIC_MB_SIZE, + .shift = 11, + .mask = 0xff +}; + +static const struct hantro_reg vp8_dec_mb_width_ext = { + .base = VDPU_REG_VP8_PIC_MB_SIZE, + .shift = 3, + .mask = 0x7 +}; + +static const struct hantro_reg vp8_dec_mb_height_ext = { + .base = VDPU_REG_VP8_PIC_MB_SIZE, + .shift = 0, + .mask = 0x7 +}; + +static const struct hantro_reg vp8_dec_bool_range = { + .base = VDPU_REG_VP8_CTRL0, + .shift = 0, + .mask = 0xff +}; + +static const struct hantro_reg vp8_dec_bool_value = { + .base = VDPU_REG_VP8_CTRL0, + .shift = 8, + .mask = 0xff +}; + +static const struct hantro_reg vp8_dec_filter_disable = { + .base = VDPU_REG_DEC_CTRL0, + .shift = 8, + .mask = 1 +}; + +static const struct hantro_reg vp8_dec_skip_mode = { + .base = VDPU_REG_DEC_CTRL0, + .shift = 9, + .mask = 1 +}; + +static const struct hantro_reg vp8_dec_start_dec = { + .base = VDPU_REG_EN_FLAGS, + .shift = 0, + .mask = 1 +}; + +static void cfg_lf(struct hantro_ctx *ctx, + const struct v4l2_ctrl_vp8_frame_header *hdr) +{ + const struct v4l2_vp8_segment_header *seg = &hdr->segment_header; + const struct v4l2_vp8_loopfilter_header *lf = &hdr->lf_header; + struct hantro_dev *vpu = ctx->dev; + unsigned int i; + u32 reg; + + if (!(seg->flags & V4L2_VP8_SEGMENT_HEADER_FLAG_ENABLED)) { + hantro_reg_write(vpu, &vp8_dec_lf_level[0], lf->level); + } else if (seg->flags & V4L2_VP8_SEGMENT_HEADER_FLAG_DELTA_VALUE_MODE) { + for (i = 0; i < 4; i++) { + u32 lf_level = clamp(lf->level + seg->lf_update[i], + 0, 63); + + hantro_reg_write(vpu, &vp8_dec_lf_level[i], lf_level); + } + } else { + for (i = 0; i < 4; i++) + hantro_reg_write(vpu, &vp8_dec_lf_level[i], + seg->lf_update[i]); + } + + reg = VDPU_REG_REF_PIC_FILT_SHARPNESS(lf->sharpness_level); + if (lf->flags & V4L2_VP8_LF_FILTER_TYPE_SIMPLE) + reg |= VDPU_REG_REF_PIC_FILT_TYPE_E; + vdpu_write_relaxed(vpu, reg, VDPU_REG_FILTER_MB_ADJ); + + if (lf->flags & V4L2_VP8_LF_HEADER_ADJ_ENABLE) { + for (i = 0; i < 4; i++) { + hantro_reg_write(vpu, &vp8_dec_mb_adj[i], + lf->mb_mode_delta[i]); + hantro_reg_write(vpu, &vp8_dec_ref_adj[i], + lf->ref_frm_delta[i]); + } + } +} + +static void cfg_qp(struct hantro_ctx *ctx, + const struct v4l2_ctrl_vp8_frame_header *hdr) +{ + const struct v4l2_vp8_quantization_header *q = &hdr->quant_header; + const struct v4l2_vp8_segment_header *seg = &hdr->segment_header; + struct hantro_dev *vpu = ctx->dev; + unsigned int i; + + if (!(seg->flags & V4L2_VP8_SEGMENT_HEADER_FLAG_ENABLED)) { + hantro_reg_write(vpu, &vp8_dec_quant[0], q->y_ac_qi); + } else if (seg->flags & V4L2_VP8_SEGMENT_HEADER_FLAG_DELTA_VALUE_MODE) { + for (i = 0; i < 4; i++) { + u32 quant = clamp(q->y_ac_qi + seg->quant_update[i], + 0, 127); + + hantro_reg_write(vpu, &vp8_dec_quant[i], quant); + } + } else { + for (i = 0; i < 4; i++) + hantro_reg_write(vpu, &vp8_dec_quant[i], + seg->quant_update[i]); + } + + hantro_reg_write(vpu, &vp8_dec_quant_delta[0], q->y_dc_delta); + hantro_reg_write(vpu, &vp8_dec_quant_delta[1], q->y2_dc_delta); + hantro_reg_write(vpu, &vp8_dec_quant_delta[2], q->y2_ac_delta); + hantro_reg_write(vpu, &vp8_dec_quant_delta[3], q->uv_dc_delta); + hantro_reg_write(vpu, &vp8_dec_quant_delta[4], q->uv_ac_delta); +} + +static void cfg_parts(struct hantro_ctx *ctx, + const struct v4l2_ctrl_vp8_frame_header *hdr) +{ + struct hantro_dev *vpu = ctx->dev; + struct vb2_v4l2_buffer *vb2_src; + u32 first_part_offset = VP8_FRAME_IS_KEY_FRAME(hdr) ? 10 : 3; + u32 mb_size, mb_offset_bytes, mb_offset_bits, mb_start_bits; + u32 dct_size_part_size, dct_part_offset; + dma_addr_t src_dma; + u32 dct_part_total_len = 0; + u32 count = 0; + unsigned int i; + + vb2_src = hantro_get_src_buf(ctx); + src_dma = vb2_dma_contig_plane_dma_addr(&vb2_src->vb2_buf, 0); + + /* + * Calculate control partition mb data info + * @first_part_header_bits: bits offset of mb data from first + * part start pos + * @mb_offset_bits: bits offset of mb data from src_dma + * base addr + * @mb_offset_byte: bytes offset of mb data from src_dma + * base addr + * @mb_start_bits: bits offset of mb data from mb data + * 64bits alignment addr + */ + mb_offset_bits = first_part_offset * 8 + + hdr->first_part_header_bits + 8; + mb_offset_bytes = mb_offset_bits / 8; + mb_start_bits = mb_offset_bits - + (mb_offset_bytes & (~DEC_8190_ALIGN_MASK)) * 8; + mb_size = hdr->first_part_size - + (mb_offset_bytes - first_part_offset) + + (mb_offset_bytes & DEC_8190_ALIGN_MASK); + + /* Macroblock data aligned base addr */ + vdpu_write_relaxed(vpu, (mb_offset_bytes & (~DEC_8190_ALIGN_MASK)) + + src_dma, VDPU_REG_VP8_ADDR_CTRL_PART); + hantro_reg_write(vpu, &vp8_dec_mb_start_bit, mb_start_bits); + hantro_reg_write(vpu, &vp8_dec_mb_aligned_data_len, mb_size); + + /* + * Calculate DCT partition info + * @dct_size_part_size: Containing sizes of DCT part, every DCT part + * has 3 bytes to store its size, except the last + * DCT part + * @dct_part_offset: bytes offset of DCT parts from src_dma base addr + * @dct_part_total_len: total size of all DCT parts + */ + dct_size_part_size = (hdr->num_dct_parts - 1) * 3; + dct_part_offset = first_part_offset + hdr->first_part_size; + for (i = 0; i < hdr->num_dct_parts; i++) + dct_part_total_len += hdr->dct_part_sizes[i]; + dct_part_total_len += dct_size_part_size; + dct_part_total_len += (dct_part_offset & DEC_8190_ALIGN_MASK); + + /* Number of DCT partitions */ + hantro_reg_write(vpu, &vp8_dec_num_dct_partitions, + hdr->num_dct_parts - 1); + + /* DCT partition length */ + hantro_reg_write(vpu, &vp8_dec_stream_len, dct_part_total_len); + + /* DCT partitions base address */ + for (i = 0; i < hdr->num_dct_parts; i++) { + u32 byte_offset = dct_part_offset + dct_size_part_size + count; + u32 base_addr = byte_offset + src_dma; + + hantro_reg_write(vpu, &vp8_dec_dct_base[i], + base_addr & (~DEC_8190_ALIGN_MASK)); + + hantro_reg_write(vpu, &vp8_dec_dct_start_bits[i], + (byte_offset & DEC_8190_ALIGN_MASK) * 8); + + count += hdr->dct_part_sizes[i]; + } +} + +/* + * prediction filter taps + * normal 6-tap filters + */ +static void cfg_tap(struct hantro_ctx *ctx, + const struct v4l2_ctrl_vp8_frame_header *hdr) +{ + struct hantro_dev *vpu = ctx->dev; + int i, j; + + if ((hdr->version & 0x03) != 0) + return; /* Tap filter not used. */ + + for (i = 0; i < 8; i++) { + for (j = 0; j < 6; j++) { + if (vp8_dec_pred_bc_tap[i][j].base != 0) + hantro_reg_write(vpu, + &vp8_dec_pred_bc_tap[i][j], + hantro_vp8_dec_mc_filter[i][j]); + } + } +} + +static void cfg_ref(struct hantro_ctx *ctx, + const struct v4l2_ctrl_vp8_frame_header *hdr) +{ + struct hantro_dev *vpu = ctx->dev; + struct vb2_v4l2_buffer *vb2_dst; + struct vb2_queue *cap_q; + dma_addr_t ref; + + cap_q = v4l2_m2m_get_vq(ctx->fh.m2m_ctx, V4L2_BUF_TYPE_VIDEO_CAPTURE); + vb2_dst = hantro_get_dst_buf(ctx); + + ref = hantro_get_ref(cap_q, hdr->last_frame_ts); + if (!ref) + ref = vb2_dma_contig_plane_dma_addr(&vb2_dst->vb2_buf, 0); + vdpu_write_relaxed(vpu, ref, VDPU_REG_VP8_ADDR_REF0); + + ref = hantro_get_ref(cap_q, hdr->golden_frame_ts); + WARN_ON(!ref && hdr->golden_frame_ts); + if (!ref) + ref = vb2_dma_contig_plane_dma_addr(&vb2_dst->vb2_buf, 0); + if (hdr->flags & V4L2_VP8_FRAME_HEADER_FLAG_SIGN_BIAS_GOLDEN) + ref |= VDPU_REG_VP8_GREF_SIGN_BIAS; + vdpu_write_relaxed(vpu, ref, VDPU_REG_VP8_ADDR_REF2_5(2)); + + ref = hantro_get_ref(cap_q, hdr->alt_frame_ts); + WARN_ON(!ref && hdr->alt_frame_ts); + if (!ref) + ref = vb2_dma_contig_plane_dma_addr(&vb2_dst->vb2_buf, 0); + if (hdr->flags & V4L2_VP8_FRAME_HEADER_FLAG_SIGN_BIAS_ALT) + ref |= VDPU_REG_VP8_AREF_SIGN_BIAS; + vdpu_write_relaxed(vpu, ref, VDPU_REG_VP8_ADDR_REF2_5(3)); +} + +static void cfg_buffers(struct hantro_ctx *ctx, + const struct v4l2_ctrl_vp8_frame_header *hdr) +{ + const struct v4l2_vp8_segment_header *seg = &hdr->segment_header; + struct hantro_dev *vpu = ctx->dev; + struct vb2_v4l2_buffer *vb2_dst; + dma_addr_t dst_dma; + u32 reg; + + vb2_dst = hantro_get_dst_buf(ctx); + + /* Set probability table buffer address */ + vdpu_write_relaxed(vpu, ctx->vp8_dec.prob_tbl.dma, + VDPU_REG_ADDR_QTABLE); + + /* Set segment map address */ + reg = VDPU_REG_FWD_PIC1_SEGMENT_BASE(ctx->vp8_dec.segment_map.dma); + if (seg->flags & V4L2_VP8_SEGMENT_HEADER_FLAG_ENABLED) { + reg |= VDPU_REG_FWD_PIC1_SEGMENT_E; + if (seg->flags & V4L2_VP8_SEGMENT_HEADER_FLAG_UPDATE_MAP) + reg |= VDPU_REG_FWD_PIC1_SEGMENT_UPD_E; + } + vdpu_write_relaxed(vpu, reg, VDPU_REG_VP8_SEGMENT_VAL); + + /* set output frame buffer address */ + dst_dma = vb2_dma_contig_plane_dma_addr(&vb2_dst->vb2_buf, 0); + vdpu_write_relaxed(vpu, dst_dma, VDPU_REG_ADDR_DST); +} + +void rk3399_vpu_vp8_dec_run(struct hantro_ctx *ctx) +{ + const struct v4l2_ctrl_vp8_frame_header *hdr; + struct hantro_dev *vpu = ctx->dev; + size_t height = ctx->dst_fmt.height; + size_t width = ctx->dst_fmt.width; + u32 mb_width, mb_height; + u32 reg; + + hantro_prepare_run(ctx); + + hdr = hantro_get_ctrl(ctx, V4L2_CID_MPEG_VIDEO_VP8_FRAME_HEADER); + if (WARN_ON(!hdr)) + return; + + /* Reset segment_map buffer in keyframe */ + if (VP8_FRAME_IS_KEY_FRAME(hdr) && ctx->vp8_dec.segment_map.cpu) + memset(ctx->vp8_dec.segment_map.cpu, 0, + ctx->vp8_dec.segment_map.size); + + hantro_vp8_prob_update(ctx, hdr); + + /* + * Extensive testing shows that the hardware does not properly + * clear the internal state from previous a decoding run. This + * causes corruption in decoded frames for multi-instance use cases. + * A soft reset before programming the registers has been found + * to resolve those problems. + */ + ctx->codec_ops->reset(ctx); + + reg = VDPU_REG_CONFIG_DEC_TIMEOUT_E + | VDPU_REG_CONFIG_DEC_CLK_GATE_E; + if (!VP8_FRAME_IS_KEY_FRAME(hdr)) + reg |= VDPU_REG_DEC_CTRL0_PIC_INTER_E; + vdpu_write_relaxed(vpu, reg, VDPU_REG_EN_FLAGS); + + reg = VDPU_REG_CONFIG_DEC_STRENDIAN_E + | VDPU_REG_CONFIG_DEC_INSWAP32_E + | VDPU_REG_CONFIG_DEC_STRSWAP32_E + | VDPU_REG_CONFIG_DEC_OUTSWAP32_E + | VDPU_REG_CONFIG_DEC_IN_ENDIAN + | VDPU_REG_CONFIG_DEC_OUT_ENDIAN; + vdpu_write_relaxed(vpu, reg, VDPU_REG_DATA_ENDIAN); + + reg = VDPU_REG_CONFIG_DEC_MAX_BURST(16); + vdpu_write_relaxed(vpu, reg, VDPU_REG_AXI_CTRL); + + reg = VDPU_REG_DEC_CTRL0_DEC_MODE(10); + vdpu_write_relaxed(vpu, reg, VDPU_REG_DEC_FORMAT); + + if (!(hdr->flags & V4L2_VP8_FRAME_HEADER_FLAG_MB_NO_SKIP_COEFF)) + hantro_reg_write(vpu, &vp8_dec_skip_mode, 1); + if (hdr->lf_header.level == 0) + hantro_reg_write(vpu, &vp8_dec_filter_disable, 1); + + /* Frame dimensions */ + mb_width = VP8_MB_WIDTH(width); + mb_height = VP8_MB_HEIGHT(height); + + hantro_reg_write(vpu, &vp8_dec_mb_width, mb_width); + hantro_reg_write(vpu, &vp8_dec_mb_height, mb_height); + hantro_reg_write(vpu, &vp8_dec_mb_width_ext, mb_width >> 9); + hantro_reg_write(vpu, &vp8_dec_mb_height_ext, mb_height >> 8); + + /* Boolean decoder */ + hantro_reg_write(vpu, &vp8_dec_bool_range, hdr->coder_state.range); + hantro_reg_write(vpu, &vp8_dec_bool_value, hdr->coder_state.value); + + reg = vdpu_read(vpu, VDPU_REG_VP8_DCT_START_BIT); + if (hdr->version != 3) + reg |= VDPU_REG_DEC_CTRL4_VC1_HEIGHT_EXT; + if (hdr->version & 0x3) + reg |= VDPU_REG_DEC_CTRL4_BILIN_MC_E; + vdpu_write_relaxed(vpu, reg, VDPU_REG_VP8_DCT_START_BIT); + + cfg_lf(ctx, hdr); + cfg_qp(ctx, hdr); + cfg_parts(ctx, hdr); + cfg_tap(ctx, hdr); + cfg_ref(ctx, hdr); + cfg_buffers(ctx, hdr); + + hantro_finish_run(ctx); + + hantro_reg_write(vpu, &vp8_dec_start_dec, 1); +} |