diff options
Diffstat (limited to 'drivers/media/platform/chips-media')
16 files changed, 1007 insertions, 532 deletions
diff --git a/drivers/media/platform/chips-media/coda/coda-bit.c b/drivers/media/platform/chips-media/coda/coda-bit.c index ed47d5bd8d61..84ded154adfe 100644 --- a/drivers/media/platform/chips-media/coda/coda-bit.c +++ b/drivers/media/platform/chips-media/coda/coda-bit.c @@ -585,7 +585,7 @@ static int coda_alloc_context_buffers(struct coda_ctx *ctx, if (!ctx->slicebuf.vaddr && q_data->fourcc == V4L2_PIX_FMT_H264) { /* worst case slice size */ - size = (DIV_ROUND_UP(q_data->rect.width, 16) * + size = (unsigned long)(DIV_ROUND_UP(q_data->rect.width, 16) * DIV_ROUND_UP(q_data->rect.height, 16)) * 3200 / 8 + 512; ret = coda_alloc_context_buf(ctx, &ctx->slicebuf, size, "slicebuf"); diff --git a/drivers/media/platform/chips-media/coda/coda-common.c b/drivers/media/platform/chips-media/coda/coda-common.c index 7da0194ec850..e6e3f5ec24f6 100644 --- a/drivers/media/platform/chips-media/coda/coda-common.c +++ b/drivers/media/platform/chips-media/coda/coda-common.c @@ -2171,8 +2171,6 @@ static const struct vb2_ops coda_qops = { .buf_queue = coda_buf_queue, .start_streaming = coda_start_streaming, .stop_streaming = coda_stop_streaming, - .wait_prepare = vb2_ops_wait_prepare, - .wait_finish = vb2_ops_wait_finish, }; static int coda_s_ctrl(struct v4l2_ctrl *ctrl) @@ -3342,11 +3340,12 @@ static int coda_runtime_resume(struct device *dev) static const struct dev_pm_ops coda_pm_ops = { SET_RUNTIME_PM_OPS(NULL, coda_runtime_resume, NULL) + SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend, pm_runtime_force_resume) }; static struct platform_driver coda_driver = { .probe = coda_probe, - .remove_new = coda_remove, + .remove = coda_remove, .driver = { .name = CODA_NAME, .of_match_table = coda_dt_ids, diff --git a/drivers/media/platform/chips-media/coda/coda-jpeg.c b/drivers/media/platform/chips-media/coda/coda-jpeg.c index ba8f41002917..5746892658b1 100644 --- a/drivers/media/platform/chips-media/coda/coda-jpeg.c +++ b/drivers/media/platform/chips-media/coda/coda-jpeg.c @@ -5,7 +5,7 @@ * Copyright (C) 2014 Philipp Zabel, Pengutronix */ -#include <asm/unaligned.h> +#include <linux/unaligned.h> #include <linux/irqreturn.h> #include <linux/kernel.h> #include <linux/ktime.h> diff --git a/drivers/media/platform/chips-media/wave5/wave5-helper.c b/drivers/media/platform/chips-media/wave5/wave5-helper.c index 8433ecab230c..2c9d8cbca6e4 100644 --- a/drivers/media/platform/chips-media/wave5/wave5-helper.c +++ b/drivers/media/platform/chips-media/wave5/wave5-helper.c @@ -7,6 +7,8 @@ #include "wave5-helper.h" +#define DEFAULT_BS_SIZE(width, height) ((width) * (height) / 8 * 3) + const char *state_to_str(enum vpu_instance_state state) { switch (state) { @@ -29,7 +31,13 @@ void wave5_cleanup_instance(struct vpu_instance *inst) { int i; - if (list_is_singular(&inst->list)) + /* + * For Wave515 SRAM memory is allocated at + * wave5_vpu_dec_register_device() and freed at + * wave5_vpu_dec_unregister_device(). + */ + if (list_is_singular(&inst->list) && + inst->dev->product_code != WAVE515_CODE) wave5_vdi_free_sram(inst->dev); for (i = 0; i < inst->fbc_buf_count; i++) @@ -52,11 +60,11 @@ int wave5_vpu_release_device(struct file *filp, char *name) { struct vpu_instance *inst = wave5_to_vpu_inst(filp->private_data); + int ret = 0; v4l2_m2m_ctx_release(inst->v4l2_fh.m2m_ctx); if (inst->state != VPU_INST_STATE_NONE) { u32 fail_res; - int ret; ret = close_func(inst, &fail_res); if (fail_res == WAVE5_SYSERR_VPU_STILL_RUNNING) { @@ -72,7 +80,7 @@ int wave5_vpu_release_device(struct file *filp, wave5_cleanup_instance(inst); - return 0; + return ret; } int wave5_vpu_queue_init(void *priv, struct vb2_queue *src_vq, struct vb2_queue *dst_vq, @@ -211,3 +219,25 @@ void wave5_return_bufs(struct vb2_queue *q, u32 state) v4l2_m2m_buf_done(vbuf, state); } } + +void wave5_update_pix_fmt(struct v4l2_pix_format_mplane *pix_mp, + int pix_fmt_type, + unsigned int width, + unsigned int height, + const struct v4l2_frmsize_stepwise *frmsize) +{ + v4l2_apply_frmsize_constraints(&width, &height, frmsize); + + if (pix_fmt_type == VPU_FMT_TYPE_CODEC) { + pix_mp->width = width; + pix_mp->height = height; + pix_mp->num_planes = 1; + pix_mp->plane_fmt[0].bytesperline = 0; + pix_mp->plane_fmt[0].sizeimage = max(DEFAULT_BS_SIZE(width, height), + pix_mp->plane_fmt[0].sizeimage); + } else { + v4l2_fill_pixfmt_mp(pix_mp, pix_mp->pixelformat, width, height); + } + pix_mp->flags = 0; + pix_mp->field = V4L2_FIELD_NONE; +} diff --git a/drivers/media/platform/chips-media/wave5/wave5-helper.h b/drivers/media/platform/chips-media/wave5/wave5-helper.h index 6cee1c14d3ce..9937fce553fc 100644 --- a/drivers/media/platform/chips-media/wave5/wave5-helper.h +++ b/drivers/media/platform/chips-media/wave5/wave5-helper.h @@ -28,4 +28,9 @@ const struct vpu_format *wave5_find_vpu_fmt_by_idx(unsigned int idx, const struct vpu_format fmt_list[MAX_FMTS]); enum wave_std wave5_to_vpu_std(unsigned int v4l2_pix_fmt, enum vpu_instance_type type); void wave5_return_bufs(struct vb2_queue *q, u32 state); +void wave5_update_pix_fmt(struct v4l2_pix_format_mplane *pix_mp, + int pix_fmt_type, + unsigned int width, + unsigned int height, + const struct v4l2_frmsize_stepwise *frmsize); #endif diff --git a/drivers/media/platform/chips-media/wave5/wave5-hw.c b/drivers/media/platform/chips-media/wave5/wave5-hw.c index 2d82791f575e..d94cf84c3ee5 100644 --- a/drivers/media/platform/chips-media/wave5/wave5-hw.c +++ b/drivers/media/platform/chips-media/wave5/wave5-hw.c @@ -18,22 +18,33 @@ #define QUEUE_REPORT_MASK 0xffff /* Encoder support fields */ -#define FEATURE_HEVC10BIT_ENC BIT(3) -#define FEATURE_AVC10BIT_ENC BIT(11) -#define FEATURE_AVC_ENCODER BIT(1) -#define FEATURE_HEVC_ENCODER BIT(0) +#define W521_FEATURE_HEVC10BIT_ENC BIT(3) +#define W521_FEATURE_AVC10BIT_ENC BIT(11) +#define W521_FEATURE_AVC_ENCODER BIT(1) +#define W521_FEATURE_HEVC_ENCODER BIT(0) + +#define ENC_AVC_INTRA_IDR_PARAM_MASK 0x7ff +#define ENC_AVC_INTRA_PERIOD_SHIFT 6 +#define ENC_AVC_IDR_PERIOD_SHIFT 17 +#define ENC_AVC_FORCED_IDR_HEADER_SHIFT 28 + +#define ENC_HEVC_INTRA_QP_SHIFT 3 +#define ENC_HEVC_FORCED_IDR_HEADER_SHIFT 9 +#define ENC_HEVC_INTRA_PERIOD_SHIFT 16 /* Decoder support fields */ -#define FEATURE_AVC_DECODER BIT(3) -#define FEATURE_HEVC_DECODER BIT(2) +#define W521_FEATURE_AVC_DECODER BIT(3) +#define W521_FEATURE_HEVC_DECODER BIT(2) +#define W515_FEATURE_HEVC10BIT_DEC BIT(1) +#define W515_FEATURE_HEVC_DECODER BIT(0) -#define FEATURE_BACKBONE BIT(16) -#define FEATURE_VCORE_BACKBONE BIT(22) -#define FEATURE_VCPU_BACKBONE BIT(28) +#define W521_FEATURE_BACKBONE BIT(16) +#define W521_FEATURE_VCORE_BACKBONE BIT(22) +#define W521_FEATURE_VCPU_BACKBONE BIT(28) #define REMAP_CTRL_MAX_SIZE_BITS ((W5_REMAP_MAX_SIZE >> 12) & 0x1ff) #define REMAP_CTRL_REGISTER_VALUE(index) ( \ - (BIT(31) | (index << 12) | BIT(11) | REMAP_CTRL_MAX_SIZE_BITS) \ + (BIT(31) | ((index) << 12) | BIT(11) | REMAP_CTRL_MAX_SIZE_BITS)\ ) #define FASTIO_ADDRESS_MASK GENMASK(15, 0) @@ -155,6 +166,8 @@ static int wave5_wait_bus_busy(struct vpu_device *vpu_dev, unsigned int addr) { u32 gdi_status_check_value = 0x3f; + if (vpu_dev->product_code == WAVE515_CODE) + gdi_status_check_value = 0x0738; if (vpu_dev->product_code == WAVE521C_CODE || vpu_dev->product_code == WAVE521_CODE || vpu_dev->product_code == WAVE521E1_CODE) @@ -186,6 +199,8 @@ unsigned int wave5_vpu_get_product_id(struct vpu_device *vpu_dev) u32 val = vpu_read_reg(vpu_dev, W5_PRODUCT_NUMBER); switch (val) { + case WAVE515_CODE: + return PRODUCT_ID_515; case WAVE521C_CODE: return PRODUCT_ID_521; case WAVE521_CODE: @@ -299,6 +314,27 @@ static int wave5_send_query(struct vpu_device *vpu_dev, struct vpu_instance *ins return wave5_vpu_firmware_command_queue_error_check(vpu_dev, NULL); } +static void setup_wave5_interrupts(struct vpu_device *vpu_dev) +{ + u32 reg_val = 0; + + if (vpu_dev->attr.support_encoders) { + /* Encoder interrupt */ + reg_val |= BIT(INT_WAVE5_ENC_SET_PARAM); + reg_val |= BIT(INT_WAVE5_ENC_PIC); + reg_val |= BIT(INT_WAVE5_BSBUF_FULL); + } + + if (vpu_dev->attr.support_decoders) { + /* Decoder interrupt */ + reg_val |= BIT(INT_WAVE5_INIT_SEQ); + reg_val |= BIT(INT_WAVE5_DEC_PIC); + reg_val |= BIT(INT_WAVE5_BSBUF_EMPTY); + } + + return vpu_write_reg(vpu_dev, W5_VPU_VINT_ENABLE, reg_val); +} + static int setup_wave5_properties(struct device *dev) { struct vpu_device *vpu_dev = dev_get_drvdata(dev); @@ -328,17 +364,35 @@ static int setup_wave5_properties(struct device *dev) hw_config_def1 = vpu_read_reg(vpu_dev, W5_RET_STD_DEF1); hw_config_feature = vpu_read_reg(vpu_dev, W5_RET_CONF_FEATURE); - p_attr->support_hevc10bit_enc = FIELD_GET(FEATURE_HEVC10BIT_ENC, hw_config_feature); - p_attr->support_avc10bit_enc = FIELD_GET(FEATURE_AVC10BIT_ENC, hw_config_feature); - - p_attr->support_decoders = FIELD_GET(FEATURE_AVC_DECODER, hw_config_def1) << STD_AVC; - p_attr->support_decoders |= FIELD_GET(FEATURE_HEVC_DECODER, hw_config_def1) << STD_HEVC; - p_attr->support_encoders = FIELD_GET(FEATURE_AVC_ENCODER, hw_config_def1) << STD_AVC; - p_attr->support_encoders |= FIELD_GET(FEATURE_HEVC_ENCODER, hw_config_def1) << STD_HEVC; - - p_attr->support_backbone = FIELD_GET(FEATURE_BACKBONE, hw_config_def0); - p_attr->support_vcpu_backbone = FIELD_GET(FEATURE_VCPU_BACKBONE, hw_config_def0); - p_attr->support_vcore_backbone = FIELD_GET(FEATURE_VCORE_BACKBONE, hw_config_def0); + if (vpu_dev->product_code == WAVE515_CODE) { + p_attr->support_hevc10bit_dec = FIELD_GET(W515_FEATURE_HEVC10BIT_DEC, + hw_config_feature); + p_attr->support_decoders = FIELD_GET(W515_FEATURE_HEVC_DECODER, + hw_config_def1) << STD_HEVC; + } else { + p_attr->support_hevc10bit_enc = FIELD_GET(W521_FEATURE_HEVC10BIT_ENC, + hw_config_feature); + p_attr->support_avc10bit_enc = FIELD_GET(W521_FEATURE_AVC10BIT_ENC, + hw_config_feature); + + p_attr->support_decoders = FIELD_GET(W521_FEATURE_AVC_DECODER, + hw_config_def1) << STD_AVC; + p_attr->support_decoders |= FIELD_GET(W521_FEATURE_HEVC_DECODER, + hw_config_def1) << STD_HEVC; + p_attr->support_encoders = FIELD_GET(W521_FEATURE_AVC_ENCODER, + hw_config_def1) << STD_AVC; + p_attr->support_encoders |= FIELD_GET(W521_FEATURE_HEVC_ENCODER, + hw_config_def1) << STD_HEVC; + + p_attr->support_backbone = FIELD_GET(W521_FEATURE_BACKBONE, + hw_config_def0); + p_attr->support_vcpu_backbone = FIELD_GET(W521_FEATURE_VCPU_BACKBONE, + hw_config_def0); + p_attr->support_vcore_backbone = FIELD_GET(W521_FEATURE_VCORE_BACKBONE, + hw_config_def0); + } + + setup_wave5_interrupts(vpu_dev); return 0; } @@ -380,12 +434,18 @@ int wave5_vpu_init(struct device *dev, u8 *fw, size_t size) common_vb = &vpu_dev->common_mem; code_base = common_vb->daddr; + + if (vpu_dev->product_code == WAVE515_CODE) + code_size = WAVE515_MAX_CODE_BUF_SIZE; + else + code_size = WAVE521_MAX_CODE_BUF_SIZE; + /* ALIGN TO 4KB */ - code_size = (WAVE5_MAX_CODE_BUF_SIZE & ~0xfff); + code_size &= ~0xfff; if (code_size < size * 2) return -EINVAL; - temp_base = common_vb->daddr + WAVE5_TEMPBUF_OFFSET; + temp_base = code_base + code_size; temp_size = WAVE5_TEMPBUF_SIZE; ret = wave5_vdi_write_memory(vpu_dev, common_vb, 0, fw, size); @@ -413,22 +473,15 @@ int wave5_vpu_init(struct device *dev, u8 *fw, size_t size) /* These register must be reset explicitly */ vpu_write_reg(vpu_dev, W5_HW_OPTION, 0); - wave5_fio_writel(vpu_dev, W5_BACKBONE_PROC_EXT_ADDR, 0); - wave5_fio_writel(vpu_dev, W5_BACKBONE_AXI_PARAM, 0); - vpu_write_reg(vpu_dev, W5_SEC_AXI_PARAM, 0); - - /* Encoder interrupt */ - reg_val = BIT(INT_WAVE5_ENC_SET_PARAM); - reg_val |= BIT(INT_WAVE5_ENC_PIC); - reg_val |= BIT(INT_WAVE5_BSBUF_FULL); - /* Decoder interrupt */ - reg_val |= BIT(INT_WAVE5_INIT_SEQ); - reg_val |= BIT(INT_WAVE5_DEC_PIC); - reg_val |= BIT(INT_WAVE5_BSBUF_EMPTY); - vpu_write_reg(vpu_dev, W5_VPU_VINT_ENABLE, reg_val); + + if (vpu_dev->product_code != WAVE515_CODE) { + wave5_fio_writel(vpu_dev, W5_BACKBONE_PROC_EXT_ADDR, 0); + wave5_fio_writel(vpu_dev, W5_BACKBONE_AXI_PARAM, 0); + vpu_write_reg(vpu_dev, W5_SEC_AXI_PARAM, 0); + } reg_val = vpu_read_reg(vpu_dev, W5_VPU_RET_VPU_CONFIG0); - if (FIELD_GET(FEATURE_BACKBONE, reg_val)) { + if (FIELD_GET(W521_FEATURE_BACKBONE, reg_val)) { reg_val = ((WAVE5_PROC_AXI_ID << 28) | (WAVE5_PRP_AXI_ID << 24) | (WAVE5_FBD_Y_AXI_ID << 20) | @@ -440,6 +493,24 @@ int wave5_vpu_init(struct device *dev, u8 *fw, size_t size) wave5_fio_writel(vpu_dev, W5_BACKBONE_PROG_AXI_ID, reg_val); } + if (vpu_dev->product_code == WAVE515_CODE) { + dma_addr_t task_buf_base; + + vpu_write_reg(vpu_dev, W5_CMD_INIT_NUM_TASK_BUF, WAVE515_COMMAND_QUEUE_DEPTH); + vpu_write_reg(vpu_dev, W5_CMD_INIT_TASK_BUF_SIZE, WAVE515_ONE_TASKBUF_SIZE); + + for (i = 0; i < WAVE515_COMMAND_QUEUE_DEPTH; i++) { + task_buf_base = temp_base + temp_size + + (i * WAVE515_ONE_TASKBUF_SIZE); + vpu_write_reg(vpu_dev, + W5_CMD_INIT_ADDR_TASK_BUF0 + (i * 4), + task_buf_base); + } + + vpu_write_reg(vpu_dev, W515_CMD_ADDR_SEC_AXI, vpu_dev->sram_buf.daddr); + vpu_write_reg(vpu_dev, W515_CMD_SEC_AXI_SIZE, vpu_dev->sram_buf.size); + } + vpu_write_reg(vpu_dev, W5_VPU_BUSY_STATUS, 1); vpu_write_reg(vpu_dev, W5_COMMAND, W5_INIT_VPU); vpu_write_reg(vpu_dev, W5_VPU_REMAP_CORE_START, 1); @@ -480,30 +551,41 @@ int wave5_vpu_build_up_dec_param(struct vpu_instance *inst, return -EINVAL; } - p_dec_info->vb_work.size = WAVE521DEC_WORKBUF_SIZE; + if (vpu_dev->product == PRODUCT_ID_515) + p_dec_info->vb_work.size = WAVE515DEC_WORKBUF_SIZE; + else + p_dec_info->vb_work.size = WAVE521DEC_WORKBUF_SIZE; + ret = wave5_vdi_allocate_dma_memory(inst->dev, &p_dec_info->vb_work); if (ret) return ret; - vpu_write_reg(inst->dev, W5_CMD_DEC_VCORE_INFO, 1); + if (inst->dev->product_code != WAVE515_CODE) + vpu_write_reg(inst->dev, W5_CMD_DEC_VCORE_INFO, 1); wave5_vdi_clear_memory(inst->dev, &p_dec_info->vb_work); vpu_write_reg(inst->dev, W5_ADDR_WORK_BASE, p_dec_info->vb_work.daddr); vpu_write_reg(inst->dev, W5_WORK_SIZE, p_dec_info->vb_work.size); - vpu_write_reg(inst->dev, W5_CMD_ADDR_SEC_AXI, vpu_dev->sram_buf.daddr); - vpu_write_reg(inst->dev, W5_CMD_SEC_AXI_SIZE, vpu_dev->sram_buf.size); + if (inst->dev->product_code != WAVE515_CODE) { + vpu_write_reg(inst->dev, W5_CMD_ADDR_SEC_AXI, vpu_dev->sram_buf.daddr); + vpu_write_reg(inst->dev, W5_CMD_SEC_AXI_SIZE, vpu_dev->sram_buf.size); + } vpu_write_reg(inst->dev, W5_CMD_DEC_BS_START_ADDR, p_dec_info->stream_buf_start_addr); vpu_write_reg(inst->dev, W5_CMD_DEC_BS_SIZE, p_dec_info->stream_buf_size); /* NOTE: SDMA reads MSB first */ vpu_write_reg(inst->dev, W5_CMD_BS_PARAM, BITSTREAM_ENDIANNESS_BIG_ENDIAN); - /* This register must be reset explicitly */ - vpu_write_reg(inst->dev, W5_CMD_EXT_ADDR, 0); - vpu_write_reg(inst->dev, W5_CMD_NUM_CQ_DEPTH_M1, (COMMAND_QUEUE_DEPTH - 1)); + if (inst->dev->product_code != WAVE515_CODE) { + /* This register must be reset explicitly */ + vpu_write_reg(inst->dev, W5_CMD_EXT_ADDR, 0); + vpu_write_reg(inst->dev, W5_CMD_NUM_CQ_DEPTH_M1, + WAVE521_COMMAND_QUEUE_DEPTH - 1); + } + vpu_write_reg(inst->dev, W5_CMD_ERR_CONCEAL, 0); ret = send_firmware_command(inst, W5_CREATE_INSTANCE, true, NULL, NULL); if (ret) { wave5_vdi_free_dma_memory(vpu_dev, &p_dec_info->vb_work); @@ -553,7 +635,7 @@ static u32 get_bitstream_options(struct dec_info *info) int wave5_vpu_dec_init_seq(struct vpu_instance *inst) { struct dec_info *p_dec_info = &inst->codec_info->dec_info; - u32 cmd_option = INIT_SEQ_NORMAL; + u32 bs_option, cmd_option = INIT_SEQ_NORMAL; u32 reg_val, fail_res; int ret; @@ -563,7 +645,13 @@ int wave5_vpu_dec_init_seq(struct vpu_instance *inst) vpu_write_reg(inst->dev, W5_BS_RD_PTR, p_dec_info->stream_rd_ptr); vpu_write_reg(inst->dev, W5_BS_WR_PTR, p_dec_info->stream_wr_ptr); - vpu_write_reg(inst->dev, W5_BS_OPTION, get_bitstream_options(p_dec_info)); + bs_option = get_bitstream_options(p_dec_info); + + /* Without RD_PTR_VALID_FLAG Wave515 ignores RD_PTR value */ + if (inst->dev->product_code == WAVE515_CODE) + bs_option |= BSOPTION_RD_PTR_VALID_FLAG; + + vpu_write_reg(inst->dev, W5_BS_OPTION, bs_option); vpu_write_reg(inst->dev, W5_COMMAND_OPTION, cmd_option); vpu_write_reg(inst->dev, W5_CMD_DEC_USER_MASK, p_dec_info->user_data_enable); @@ -629,10 +717,12 @@ static void wave5_get_dec_seq_result(struct vpu_instance *inst, struct dec_initi info->profile = FIELD_GET(SEQ_PARAM_PROFILE_MASK, reg_val); } - info->vlc_buf_size = vpu_read_reg(inst->dev, W5_RET_VLC_BUF_SIZE); - info->param_buf_size = vpu_read_reg(inst->dev, W5_RET_PARAM_BUF_SIZE); - p_dec_info->vlc_buf_size = info->vlc_buf_size; - p_dec_info->param_buf_size = info->param_buf_size; + if (inst->dev->product_code != WAVE515_CODE) { + info->vlc_buf_size = vpu_read_reg(inst->dev, W5_RET_VLC_BUF_SIZE); + info->param_buf_size = vpu_read_reg(inst->dev, W5_RET_PARAM_BUF_SIZE); + p_dec_info->vlc_buf_size = info->vlc_buf_size; + p_dec_info->param_buf_size = info->param_buf_size; + } } int wave5_vpu_dec_get_seq_info(struct vpu_instance *inst, struct dec_initial_info *info) @@ -734,22 +824,27 @@ int wave5_vpu_dec_register_framebuffer(struct vpu_instance *inst, struct frame_b pic_size = (init_info->pic_width << 16) | (init_info->pic_height); - vb_buf.size = (p_dec_info->vlc_buf_size * VLC_BUF_NUM) + - (p_dec_info->param_buf_size * COMMAND_QUEUE_DEPTH); - vb_buf.daddr = 0; + if (inst->dev->product_code != WAVE515_CODE) { + vb_buf.size = (p_dec_info->vlc_buf_size * VLC_BUF_NUM) + + (p_dec_info->param_buf_size * WAVE521_COMMAND_QUEUE_DEPTH); + vb_buf.daddr = 0; - if (vb_buf.size != p_dec_info->vb_task.size) { - wave5_vdi_free_dma_memory(inst->dev, &p_dec_info->vb_task); - ret = wave5_vdi_allocate_dma_memory(inst->dev, &vb_buf); - if (ret) - goto free_fbc_c_tbl_buffers; + if (vb_buf.size != p_dec_info->vb_task.size) { + wave5_vdi_free_dma_memory(inst->dev, + &p_dec_info->vb_task); + ret = wave5_vdi_allocate_dma_memory(inst->dev, + &vb_buf); + if (ret) + goto free_fbc_c_tbl_buffers; - p_dec_info->vb_task = vb_buf; - } + p_dec_info->vb_task = vb_buf; + } - vpu_write_reg(inst->dev, W5_CMD_SET_FB_ADDR_TASK_BUF, - p_dec_info->vb_task.daddr); - vpu_write_reg(inst->dev, W5_CMD_SET_FB_TASK_BUF_SIZE, vb_buf.size); + vpu_write_reg(inst->dev, W5_CMD_SET_FB_ADDR_TASK_BUF, + p_dec_info->vb_task.daddr); + vpu_write_reg(inst->dev, W5_CMD_SET_FB_TASK_BUF_SIZE, + vb_buf.size); + } } else { pic_size = (init_info->pic_width << 16) | (init_info->pic_height); @@ -830,6 +925,43 @@ free_mv_buffers: return ret; } +static u32 wave5_vpu_dec_validate_sec_axi(struct vpu_instance *inst) +{ + u32 bitdepth = inst->codec_info->dec_info.initial_info.luma_bitdepth; + struct dec_info *p_dec_info = &inst->codec_info->dec_info; + u32 bit_size = 0, ip_size = 0, lf_size = 0, ret = 0; + u32 sram_size = inst->dev->sram_size; + u32 width = inst->src_fmt.width; + + if (!sram_size) + return 0; + + /* + * TODO: calculate bit_size, ip_size, lf_size from width and bitdepth + * for Wave521. + */ + if (inst->dev->product_code == WAVE515_CODE) { + bit_size = DIV_ROUND_UP(width, 16) * 5 * 8; + ip_size = ALIGN(width, 16) * 2 * bitdepth / 8; + lf_size = ALIGN(width, 16) * 10 * bitdepth / 8; + } + + if (p_dec_info->sec_axi_info.use_bit_enable && sram_size >= bit_size) { + ret |= BIT(0); + sram_size -= bit_size; + } + + if (p_dec_info->sec_axi_info.use_ip_enable && sram_size >= ip_size) { + ret |= BIT(9); + sram_size -= ip_size; + } + + if (p_dec_info->sec_axi_info.use_lf_row_enable && sram_size >= lf_size) + ret |= BIT(15); + + return ret; +} + int wave5_vpu_decode(struct vpu_instance *inst, u32 *fail_res) { u32 reg_val; @@ -842,9 +974,7 @@ int wave5_vpu_decode(struct vpu_instance *inst, u32 *fail_res) vpu_write_reg(inst->dev, W5_BS_OPTION, get_bitstream_options(p_dec_info)); /* secondary AXI */ - reg_val = p_dec_info->sec_axi_info.use_bit_enable | - (p_dec_info->sec_axi_info.use_ip_enable << 9) | - (p_dec_info->sec_axi_info.use_lf_row_enable << 15); + reg_val = wave5_vpu_dec_validate_sec_axi(inst); vpu_write_reg(inst->dev, W5_USE_SEC_AXI, reg_val); /* set attributes of user buffer */ @@ -992,11 +1122,18 @@ int wave5_vpu_re_init(struct device *dev, u8 *fw, size_t size) common_vb = &vpu_dev->common_mem; code_base = common_vb->daddr; + + if (vpu_dev->product_code == WAVE515_CODE) + code_size = WAVE515_MAX_CODE_BUF_SIZE; + else + code_size = WAVE521_MAX_CODE_BUF_SIZE; + /* ALIGN TO 4KB */ - code_size = (WAVE5_MAX_CODE_BUF_SIZE & ~0xfff); + code_size &= ~0xfff; if (code_size < size * 2) return -EINVAL; - temp_base = common_vb->daddr + WAVE5_TEMPBUF_OFFSET; + + temp_base = code_base + code_size; temp_size = WAVE5_TEMPBUF_SIZE; old_code_base = vpu_read_reg(vpu_dev, W5_VPU_REMAP_PADDR); @@ -1030,22 +1167,15 @@ int wave5_vpu_re_init(struct device *dev, u8 *fw, size_t size) /* These register must be reset explicitly */ vpu_write_reg(vpu_dev, W5_HW_OPTION, 0); - wave5_fio_writel(vpu_dev, W5_BACKBONE_PROC_EXT_ADDR, 0); - wave5_fio_writel(vpu_dev, W5_BACKBONE_AXI_PARAM, 0); - vpu_write_reg(vpu_dev, W5_SEC_AXI_PARAM, 0); - /* Encoder interrupt */ - reg_val = BIT(INT_WAVE5_ENC_SET_PARAM); - reg_val |= BIT(INT_WAVE5_ENC_PIC); - reg_val |= BIT(INT_WAVE5_BSBUF_FULL); - /* Decoder interrupt */ - reg_val |= BIT(INT_WAVE5_INIT_SEQ); - reg_val |= BIT(INT_WAVE5_DEC_PIC); - reg_val |= BIT(INT_WAVE5_BSBUF_EMPTY); - vpu_write_reg(vpu_dev, W5_VPU_VINT_ENABLE, reg_val); + if (vpu_dev->product_code != WAVE515_CODE) { + wave5_fio_writel(vpu_dev, W5_BACKBONE_PROC_EXT_ADDR, 0); + wave5_fio_writel(vpu_dev, W5_BACKBONE_AXI_PARAM, 0); + vpu_write_reg(vpu_dev, W5_SEC_AXI_PARAM, 0); + } reg_val = vpu_read_reg(vpu_dev, W5_VPU_RET_VPU_CONFIG0); - if (FIELD_GET(FEATURE_BACKBONE, reg_val)) { + if (FIELD_GET(W521_FEATURE_BACKBONE, reg_val)) { reg_val = ((WAVE5_PROC_AXI_ID << 28) | (WAVE5_PRP_AXI_ID << 24) | (WAVE5_FBD_Y_AXI_ID << 20) | @@ -1057,6 +1187,29 @@ int wave5_vpu_re_init(struct device *dev, u8 *fw, size_t size) wave5_fio_writel(vpu_dev, W5_BACKBONE_PROG_AXI_ID, reg_val); } + if (vpu_dev->product_code == WAVE515_CODE) { + dma_addr_t task_buf_base; + u32 i; + + vpu_write_reg(vpu_dev, W5_CMD_INIT_NUM_TASK_BUF, + WAVE515_COMMAND_QUEUE_DEPTH); + vpu_write_reg(vpu_dev, W5_CMD_INIT_TASK_BUF_SIZE, + WAVE515_ONE_TASKBUF_SIZE); + + for (i = 0; i < WAVE515_COMMAND_QUEUE_DEPTH; i++) { + task_buf_base = temp_base + temp_size + + (i * WAVE515_ONE_TASKBUF_SIZE); + vpu_write_reg(vpu_dev, + W5_CMD_INIT_ADDR_TASK_BUF0 + (i * 4), + task_buf_base); + } + + vpu_write_reg(vpu_dev, W515_CMD_ADDR_SEC_AXI, + vpu_dev->sram_buf.daddr); + vpu_write_reg(vpu_dev, W515_CMD_SEC_AXI_SIZE, + vpu_dev->sram_buf.size); + } + vpu_write_reg(vpu_dev, W5_VPU_BUSY_STATUS, 1); vpu_write_reg(vpu_dev, W5_COMMAND, W5_INIT_VPU); vpu_write_reg(vpu_dev, W5_VPU_REMAP_CORE_START, 1); @@ -1075,13 +1228,13 @@ int wave5_vpu_re_init(struct device *dev, u8 *fw, size_t size) return setup_wave5_properties(dev); } -static int wave5_vpu_sleep_wake(struct device *dev, bool i_sleep_wake, const uint16_t *code, - size_t size) +int wave5_vpu_sleep_wake(struct device *dev, bool i_sleep_wake, const uint16_t *code, + size_t size) { u32 reg_val; struct vpu_buf *common_vb; - dma_addr_t code_base; - u32 code_size, reason_code; + dma_addr_t code_base, temp_base; + u32 code_size, temp_size, reason_code; struct vpu_device *vpu_dev = dev_get_drvdata(dev); int ret; @@ -1111,13 +1264,22 @@ static int wave5_vpu_sleep_wake(struct device *dev, bool i_sleep_wake, const uin common_vb = &vpu_dev->common_mem; code_base = common_vb->daddr; + + if (vpu_dev->product_code == WAVE515_CODE) + code_size = WAVE515_MAX_CODE_BUF_SIZE; + else + code_size = WAVE521_MAX_CODE_BUF_SIZE; + /* ALIGN TO 4KB */ - code_size = (WAVE5_MAX_CODE_BUF_SIZE & ~0xfff); + code_size &= ~0xfff; if (code_size < size * 2) { dev_err(dev, "size too small\n"); return -EINVAL; } + temp_base = code_base + code_size; + temp_size = WAVE5_TEMPBUF_SIZE; + /* Power on without DEBUG mode */ vpu_write_reg(vpu_dev, W5_PO_CONF, 0); @@ -1130,22 +1292,17 @@ static int wave5_vpu_sleep_wake(struct device *dev, bool i_sleep_wake, const uin /* These register must be reset explicitly */ vpu_write_reg(vpu_dev, W5_HW_OPTION, 0); - wave5_fio_writel(vpu_dev, W5_BACKBONE_PROC_EXT_ADDR, 0); - wave5_fio_writel(vpu_dev, W5_BACKBONE_AXI_PARAM, 0); - vpu_write_reg(vpu_dev, W5_SEC_AXI_PARAM, 0); - /* Encoder interrupt */ - reg_val = BIT(INT_WAVE5_ENC_SET_PARAM); - reg_val |= BIT(INT_WAVE5_ENC_PIC); - reg_val |= BIT(INT_WAVE5_BSBUF_FULL); - /* Decoder interrupt */ - reg_val |= BIT(INT_WAVE5_INIT_SEQ); - reg_val |= BIT(INT_WAVE5_DEC_PIC); - reg_val |= BIT(INT_WAVE5_BSBUF_EMPTY); - vpu_write_reg(vpu_dev, W5_VPU_VINT_ENABLE, reg_val); + if (vpu_dev->product_code != WAVE515_CODE) { + wave5_fio_writel(vpu_dev, W5_BACKBONE_PROC_EXT_ADDR, 0); + wave5_fio_writel(vpu_dev, W5_BACKBONE_AXI_PARAM, 0); + vpu_write_reg(vpu_dev, W5_SEC_AXI_PARAM, 0); + } + + setup_wave5_interrupts(vpu_dev); reg_val = vpu_read_reg(vpu_dev, W5_VPU_RET_VPU_CONFIG0); - if (FIELD_GET(FEATURE_BACKBONE, reg_val)) { + if (FIELD_GET(W521_FEATURE_BACKBONE, reg_val)) { reg_val = ((WAVE5_PROC_AXI_ID << 28) | (WAVE5_PRP_AXI_ID << 24) | (WAVE5_FBD_Y_AXI_ID << 20) | @@ -1157,6 +1314,29 @@ static int wave5_vpu_sleep_wake(struct device *dev, bool i_sleep_wake, const uin wave5_fio_writel(vpu_dev, W5_BACKBONE_PROG_AXI_ID, reg_val); } + if (vpu_dev->product_code == WAVE515_CODE) { + dma_addr_t task_buf_base; + u32 i; + + vpu_write_reg(vpu_dev, W5_CMD_INIT_NUM_TASK_BUF, + WAVE515_COMMAND_QUEUE_DEPTH); + vpu_write_reg(vpu_dev, W5_CMD_INIT_TASK_BUF_SIZE, + WAVE515_ONE_TASKBUF_SIZE); + + for (i = 0; i < WAVE515_COMMAND_QUEUE_DEPTH; i++) { + task_buf_base = temp_base + temp_size + + (i * WAVE515_ONE_TASKBUF_SIZE); + vpu_write_reg(vpu_dev, + W5_CMD_INIT_ADDR_TASK_BUF0 + (i * 4), + task_buf_base); + } + + vpu_write_reg(vpu_dev, W515_CMD_ADDR_SEC_AXI, + vpu_dev->sram_buf.daddr); + vpu_write_reg(vpu_dev, W515_CMD_SEC_AXI_SIZE, + vpu_dev->sram_buf.size); + } + vpu_write_reg(vpu_dev, W5_VPU_BUSY_STATUS, 1); vpu_write_reg(vpu_dev, W5_COMMAND, W5_WAKEUP_VPU); /* Start VPU after settings */ @@ -1401,7 +1581,7 @@ int wave5_vpu_build_up_enc_param(struct device *dev, struct vpu_instance *inst, reg_val = (open_param->line_buf_int_en << 6) | BITSTREAM_ENDIANNESS_BIG_ENDIAN; vpu_write_reg(inst->dev, W5_CMD_BS_PARAM, reg_val); vpu_write_reg(inst->dev, W5_CMD_EXT_ADDR, 0); - vpu_write_reg(inst->dev, W5_CMD_NUM_CQ_DEPTH_M1, (COMMAND_QUEUE_DEPTH - 1)); + vpu_write_reg(inst->dev, W5_CMD_NUM_CQ_DEPTH_M1, WAVE521_COMMAND_QUEUE_DEPTH - 1); /* This register must be reset explicitly */ vpu_write_reg(inst->dev, W5_CMD_ENC_SRC_OPTIONS, 0); @@ -1601,12 +1781,19 @@ int wave5_vpu_enc_init_seq(struct vpu_instance *inst) if (inst->std == W_AVC_ENC) vpu_write_reg(inst->dev, W5_CMD_ENC_SEQ_INTRA_PARAM, p_param->intra_qp | - ((p_param->intra_period & 0x7ff) << 6) | - ((p_param->avc_idr_period & 0x7ff) << 17)); + ((p_param->intra_period & ENC_AVC_INTRA_IDR_PARAM_MASK) + << ENC_AVC_INTRA_PERIOD_SHIFT) | + ((p_param->avc_idr_period & ENC_AVC_INTRA_IDR_PARAM_MASK) + << ENC_AVC_IDR_PERIOD_SHIFT) | + (p_param->forced_idr_header_enable + << ENC_AVC_FORCED_IDR_HEADER_SHIFT)); else if (inst->std == W_HEVC_ENC) vpu_write_reg(inst->dev, W5_CMD_ENC_SEQ_INTRA_PARAM, - p_param->decoding_refresh_type | (p_param->intra_qp << 3) | - (p_param->intra_period << 16)); + p_param->decoding_refresh_type | + (p_param->intra_qp << ENC_HEVC_INTRA_QP_SHIFT) | + (p_param->forced_idr_header_enable + << ENC_HEVC_FORCED_IDR_HEADER_SHIFT) | + (p_param->intra_period << ENC_HEVC_INTRA_PERIOD_SHIFT)); reg_val = (p_param->rdo_skip << 2) | (p_param->lambda_scaling_enable << 3) | @@ -1855,7 +2042,7 @@ int wave5_vpu_enc_register_framebuffer(struct device *dev, struct vpu_instance * p_enc_info->vb_sub_sam_buf = vb_sub_sam_buf; vb_task.size = (p_enc_info->vlc_buf_size * VLC_BUF_NUM) + - (p_enc_info->param_buf_size * COMMAND_QUEUE_DEPTH); + (p_enc_info->param_buf_size * WAVE521_COMMAND_QUEUE_DEPTH); vb_task.daddr = 0; if (p_enc_info->vb_task.size == 0) { ret = wave5_vdi_allocate_dma_memory(vpu_dev, &vb_task); @@ -1943,6 +2130,31 @@ free_vb_fbc_y_tbl: return ret; } +static u32 wave5_vpu_enc_validate_sec_axi(struct vpu_instance *inst) +{ + struct enc_info *p_enc_info = &inst->codec_info->enc_info; + u32 rdo_size = 0, lf_size = 0, ret = 0; + u32 sram_size = inst->dev->sram_size; + + if (!sram_size) + return 0; + + /* + * TODO: calculate rdo_size and lf_size from inst->src_fmt.width and + * inst->codec_info->enc_info.open_param.wave_param.internal_bit_depth + */ + + if (p_enc_info->sec_axi_info.use_enc_rdo_enable && sram_size >= rdo_size) { + ret |= BIT(11); + sram_size -= rdo_size; + } + + if (p_enc_info->sec_axi_info.use_enc_lf_enable && sram_size >= lf_size) + ret |= BIT(15); + + return ret; +} + int wave5_vpu_encode(struct vpu_instance *inst, struct enc_param *option, u32 *fail_res) { u32 src_frame_format; @@ -1964,8 +2176,7 @@ int wave5_vpu_encode(struct vpu_instance *inst, struct enc_param *option, u32 *f vpu_write_reg(inst->dev, W5_CMD_ENC_PIC_SRC_AXI_SEL, DEFAULT_SRC_AXI); /* secondary AXI */ - reg_val = (p_enc_info->sec_axi_info.use_enc_rdo_enable << 11) | - (p_enc_info->sec_axi_info.use_enc_lf_enable << 15); + reg_val = wave5_vpu_enc_validate_sec_axi(inst); vpu_write_reg(inst->dev, W5_CMD_ENC_PIC_USE_SEC_AXI, reg_val); vpu_write_reg(inst->dev, W5_CMD_ENC_PIC_REPORT_PARAM, 0); diff --git a/drivers/media/platform/chips-media/wave5/wave5-regdefine.h b/drivers/media/platform/chips-media/wave5/wave5-regdefine.h index a15c6b2c3d8b..557344754c4c 100644 --- a/drivers/media/platform/chips-media/wave5/wave5-regdefine.h +++ b/drivers/media/platform/chips-media/wave5/wave5-regdefine.h @@ -205,6 +205,9 @@ enum query_opt { #define W5_ADDR_TEMP_BASE (W5_REG_BASE + 0x011C) #define W5_TEMP_SIZE (W5_REG_BASE + 0x0120) #define W5_HW_OPTION (W5_REG_BASE + 0x012C) +#define W5_CMD_INIT_NUM_TASK_BUF (W5_REG_BASE + 0x0134) +#define W5_CMD_INIT_ADDR_TASK_BUF0 (W5_REG_BASE + 0x0138) +#define W5_CMD_INIT_TASK_BUF_SIZE (W5_REG_BASE + 0x0178) #define W5_SEC_AXI_PARAM (W5_REG_BASE + 0x0180) /************************************************************************/ @@ -216,7 +219,9 @@ enum query_opt { #define W5_CMD_DEC_BS_SIZE (W5_REG_BASE + 0x0120) #define W5_CMD_BS_PARAM (W5_REG_BASE + 0x0124) #define W5_CMD_ADDR_SEC_AXI (W5_REG_BASE + 0x0130) +#define W515_CMD_ADDR_SEC_AXI (W5_REG_BASE + 0x0124) #define W5_CMD_SEC_AXI_SIZE (W5_REG_BASE + 0x0134) +#define W515_CMD_SEC_AXI_SIZE (W5_REG_BASE + 0x0128) #define W5_CMD_EXT_ADDR (W5_REG_BASE + 0x0138) #define W5_CMD_NUM_CQ_DEPTH_M1 (W5_REG_BASE + 0x013C) #define W5_CMD_ERR_CONCEAL (W5_REG_BASE + 0x0140) diff --git a/drivers/media/platform/chips-media/wave5/wave5-vdi.c b/drivers/media/platform/chips-media/wave5/wave5-vdi.c index 3809f70bc0b4..bb13267ced38 100644 --- a/drivers/media/platform/chips-media/wave5/wave5-vdi.c +++ b/drivers/media/platform/chips-media/wave5/wave5-vdi.c @@ -18,7 +18,11 @@ static int wave5_vdi_allocate_common_memory(struct device *dev) if (!vpu_dev->common_mem.vaddr) { int ret; - vpu_dev->common_mem.size = SIZE_COMMON; + if (vpu_dev->product_code == WAVE515_CODE) + vpu_dev->common_mem.size = WAVE515_SIZE_COMMON; + else + vpu_dev->common_mem.size = WAVE521_SIZE_COMMON; + ret = wave5_vdi_allocate_dma_memory(vpu_dev, &vpu_dev->common_mem); if (ret) { dev_err(dev, "unable to allocate common buffer\n"); @@ -174,16 +178,19 @@ int wave5_vdi_allocate_array(struct vpu_device *vpu_dev, struct vpu_buf *array, void wave5_vdi_allocate_sram(struct vpu_device *vpu_dev) { struct vpu_buf *vb = &vpu_dev->sram_buf; + dma_addr_t daddr; + void *vaddr; + size_t size; - if (!vpu_dev->sram_pool || !vpu_dev->sram_size) + if (!vpu_dev->sram_pool || vb->vaddr) return; - if (!vb->vaddr) { - vb->size = vpu_dev->sram_size; - vb->vaddr = gen_pool_dma_alloc(vpu_dev->sram_pool, vb->size, - &vb->daddr); - if (!vb->vaddr) - vb->size = 0; + size = min_t(size_t, vpu_dev->sram_size, gen_pool_avail(vpu_dev->sram_pool)); + vaddr = gen_pool_dma_alloc(vpu_dev->sram_pool, size, &daddr); + if (vaddr) { + vb->vaddr = vaddr; + vb->daddr = daddr; + vb->size = size; } dev_dbg(vpu_dev->dev, "%s: sram daddr: %pad, size: %zu, vaddr: 0x%p\n", @@ -197,9 +204,7 @@ void wave5_vdi_free_sram(struct vpu_device *vpu_dev) if (!vb->size || !vb->vaddr) return; - if (vb->vaddr) - gen_pool_free(vpu_dev->sram_pool, (unsigned long)vb->vaddr, - vb->size); + gen_pool_free(vpu_dev->sram_pool, (unsigned long)vb->vaddr, vb->size); memset(vb, 0, sizeof(*vb)); } diff --git a/drivers/media/platform/chips-media/wave5/wave5-vpu-dec.c b/drivers/media/platform/chips-media/wave5/wave5-vpu-dec.c index ef227af72348..fd71f0c43ac3 100644 --- a/drivers/media/platform/chips-media/wave5/wave5-vpu-dec.c +++ b/drivers/media/platform/chips-media/wave5/wave5-vpu-dec.c @@ -5,116 +5,98 @@ * Copyright (C) 2021-2023 CHIPS&MEDIA INC */ +#include <linux/pm_runtime.h> #include "wave5-helper.h" #define VPU_DEC_DEV_NAME "C&M Wave5 VPU decoder" #define VPU_DEC_DRV_NAME "wave5-dec" -#define DEFAULT_SRC_SIZE(width, height) ({ \ - (width) * (height) / 8 * 3; \ -}) +static const struct v4l2_frmsize_stepwise dec_hevc_frmsize = { + .min_width = W5_MIN_DEC_PIC_8_WIDTH, + .max_width = W5_MAX_DEC_PIC_WIDTH, + .step_width = W5_DEC_CODEC_STEP_WIDTH, + .min_height = W5_MIN_DEC_PIC_8_HEIGHT, + .max_height = W5_MAX_DEC_PIC_HEIGHT, + .step_height = W5_DEC_CODEC_STEP_HEIGHT, +}; + +static const struct v4l2_frmsize_stepwise dec_h264_frmsize = { + .min_width = W5_MIN_DEC_PIC_32_WIDTH, + .max_width = W5_MAX_DEC_PIC_WIDTH, + .step_width = W5_DEC_CODEC_STEP_WIDTH, + .min_height = W5_MIN_DEC_PIC_32_HEIGHT, + .max_height = W5_MAX_DEC_PIC_HEIGHT, + .step_height = W5_DEC_CODEC_STEP_HEIGHT, +}; + +static const struct v4l2_frmsize_stepwise dec_raw_frmsize = { + .min_width = W5_MIN_DEC_PIC_8_WIDTH, + .max_width = W5_MAX_DEC_PIC_WIDTH, + .step_width = W5_DEC_RAW_STEP_WIDTH, + .min_height = W5_MIN_DEC_PIC_8_HEIGHT, + .max_height = W5_MAX_DEC_PIC_HEIGHT, + .step_height = W5_DEC_RAW_STEP_HEIGHT, +}; static const struct vpu_format dec_fmt_list[FMT_TYPES][MAX_FMTS] = { [VPU_FMT_TYPE_CODEC] = { { .v4l2_pix_fmt = V4L2_PIX_FMT_HEVC, - .max_width = 8192, - .min_width = 8, - .max_height = 4320, - .min_height = 8, + .v4l2_frmsize = &dec_hevc_frmsize, }, { .v4l2_pix_fmt = V4L2_PIX_FMT_H264, - .max_width = 8192, - .min_width = 32, - .max_height = 4320, - .min_height = 32, + .v4l2_frmsize = &dec_h264_frmsize, }, }, [VPU_FMT_TYPE_RAW] = { { .v4l2_pix_fmt = V4L2_PIX_FMT_YUV420, - .max_width = 8192, - .min_width = 8, - .max_height = 4320, - .min_height = 8, + .v4l2_frmsize = &dec_raw_frmsize, }, { .v4l2_pix_fmt = V4L2_PIX_FMT_NV12, - .max_width = 8192, - .min_width = 8, - .max_height = 4320, - .min_height = 8, + .v4l2_frmsize = &dec_raw_frmsize, }, { .v4l2_pix_fmt = V4L2_PIX_FMT_NV21, - .max_width = 8192, - .min_width = 8, - .max_height = 4320, - .min_height = 8, + .v4l2_frmsize = &dec_raw_frmsize, }, { .v4l2_pix_fmt = V4L2_PIX_FMT_YUV422P, - .max_width = 8192, - .min_width = 8, - .max_height = 4320, - .min_height = 8, + .v4l2_frmsize = &dec_raw_frmsize, }, { .v4l2_pix_fmt = V4L2_PIX_FMT_NV16, - .max_width = 8192, - .min_width = 8, - .max_height = 4320, - .min_height = 8, + .v4l2_frmsize = &dec_raw_frmsize, }, { .v4l2_pix_fmt = V4L2_PIX_FMT_NV61, - .max_width = 8192, - .min_width = 8, - .max_height = 4320, - .min_height = 8, + .v4l2_frmsize = &dec_raw_frmsize, }, { .v4l2_pix_fmt = V4L2_PIX_FMT_YUV420M, - .max_width = 8192, - .min_width = 8, - .max_height = 4320, - .min_height = 8, + .v4l2_frmsize = &dec_raw_frmsize, }, { .v4l2_pix_fmt = V4L2_PIX_FMT_NV12M, - .max_width = 8192, - .min_width = 8, - .max_height = 4320, - .min_height = 8, + .v4l2_frmsize = &dec_raw_frmsize, }, { .v4l2_pix_fmt = V4L2_PIX_FMT_NV21M, - .max_width = 8192, - .min_width = 8, - .max_height = 4320, - .min_height = 8, + .v4l2_frmsize = &dec_raw_frmsize, }, { .v4l2_pix_fmt = V4L2_PIX_FMT_YUV422M, - .max_width = 8192, - .min_width = 8, - .max_height = 4320, - .min_height = 8, + .v4l2_frmsize = &dec_raw_frmsize, }, { .v4l2_pix_fmt = V4L2_PIX_FMT_NV16M, - .max_width = 8192, - .min_width = 8, - .max_height = 4320, - .min_height = 8, + .v4l2_frmsize = &dec_raw_frmsize, }, { .v4l2_pix_fmt = V4L2_PIX_FMT_NV61M, - .max_width = 8192, - .min_width = 8, - .max_height = 4320, - .min_height = 8, + .v4l2_frmsize = &dec_raw_frmsize, }, } }; @@ -233,74 +215,6 @@ static void wave5_handle_src_buffer(struct vpu_instance *inst, dma_addr_t rd_ptr inst->remaining_consumed_bytes = consumed_bytes; } -static void wave5_update_pix_fmt(struct v4l2_pix_format_mplane *pix_mp, unsigned int width, - unsigned int height) -{ - switch (pix_mp->pixelformat) { - case V4L2_PIX_FMT_YUV420: - case V4L2_PIX_FMT_NV12: - case V4L2_PIX_FMT_NV21: - pix_mp->width = round_up(width, 32); - pix_mp->height = round_up(height, 16); - pix_mp->plane_fmt[0].bytesperline = round_up(width, 32); - pix_mp->plane_fmt[0].sizeimage = width * height * 3 / 2; - break; - case V4L2_PIX_FMT_YUV422P: - case V4L2_PIX_FMT_NV16: - case V4L2_PIX_FMT_NV61: - pix_mp->width = round_up(width, 32); - pix_mp->height = round_up(height, 16); - pix_mp->plane_fmt[0].bytesperline = round_up(width, 32); - pix_mp->plane_fmt[0].sizeimage = width * height * 2; - break; - case V4L2_PIX_FMT_YUV420M: - pix_mp->width = round_up(width, 32); - pix_mp->height = round_up(height, 16); - pix_mp->plane_fmt[0].bytesperline = round_up(width, 32); - pix_mp->plane_fmt[0].sizeimage = width * height; - pix_mp->plane_fmt[1].bytesperline = round_up(width, 32) / 2; - pix_mp->plane_fmt[1].sizeimage = width * height / 4; - pix_mp->plane_fmt[2].bytesperline = round_up(width, 32) / 2; - pix_mp->plane_fmt[2].sizeimage = width * height / 4; - break; - case V4L2_PIX_FMT_NV12M: - case V4L2_PIX_FMT_NV21M: - pix_mp->width = round_up(width, 32); - pix_mp->height = round_up(height, 16); - pix_mp->plane_fmt[0].bytesperline = round_up(width, 32); - pix_mp->plane_fmt[0].sizeimage = width * height; - pix_mp->plane_fmt[1].bytesperline = round_up(width, 32); - pix_mp->plane_fmt[1].sizeimage = width * height / 2; - break; - case V4L2_PIX_FMT_YUV422M: - pix_mp->width = round_up(width, 32); - pix_mp->height = round_up(height, 16); - pix_mp->plane_fmt[0].bytesperline = round_up(width, 32); - pix_mp->plane_fmt[0].sizeimage = width * height; - pix_mp->plane_fmt[1].bytesperline = round_up(width, 32) / 2; - pix_mp->plane_fmt[1].sizeimage = width * height / 2; - pix_mp->plane_fmt[2].bytesperline = round_up(width, 32) / 2; - pix_mp->plane_fmt[2].sizeimage = width * height / 2; - break; - case V4L2_PIX_FMT_NV16M: - case V4L2_PIX_FMT_NV61M: - pix_mp->width = round_up(width, 32); - pix_mp->height = round_up(height, 16); - pix_mp->plane_fmt[0].bytesperline = round_up(width, 32); - pix_mp->plane_fmt[0].sizeimage = width * height; - pix_mp->plane_fmt[1].bytesperline = round_up(width, 32); - pix_mp->plane_fmt[1].sizeimage = width * height; - break; - default: - pix_mp->width = width; - pix_mp->height = height; - pix_mp->plane_fmt[0].bytesperline = 0; - pix_mp->plane_fmt[0].sizeimage = max(DEFAULT_SRC_SIZE(width, height), - pix_mp->plane_fmt[0].sizeimage); - break; - } -} - static int start_decode(struct vpu_instance *inst, u32 *fail_res) { struct v4l2_m2m_ctx *m2m_ctx = inst->v4l2_fh.m2m_ctx; @@ -388,6 +302,8 @@ static int handle_dynamic_resolution_change(struct vpu_instance *inst) } if (p_dec_info->initial_info_obtained) { + const struct vpu_format *vpu_fmt; + inst->conf_win.left = initial_info->pic_crop_rect.left; inst->conf_win.top = initial_info->pic_crop_rect.top; inst->conf_win.width = initial_info->pic_width - @@ -395,10 +311,27 @@ static int handle_dynamic_resolution_change(struct vpu_instance *inst) inst->conf_win.height = initial_info->pic_height - initial_info->pic_crop_rect.top - initial_info->pic_crop_rect.bottom; - wave5_update_pix_fmt(&inst->src_fmt, initial_info->pic_width, - initial_info->pic_height); - wave5_update_pix_fmt(&inst->dst_fmt, initial_info->pic_width, - initial_info->pic_height); + vpu_fmt = wave5_find_vpu_fmt(inst->src_fmt.pixelformat, + dec_fmt_list[VPU_FMT_TYPE_CODEC]); + if (!vpu_fmt) + return -EINVAL; + + wave5_update_pix_fmt(&inst->src_fmt, + VPU_FMT_TYPE_CODEC, + initial_info->pic_width, + initial_info->pic_height, + vpu_fmt->v4l2_frmsize); + + vpu_fmt = wave5_find_vpu_fmt(inst->dst_fmt.pixelformat, + dec_fmt_list[VPU_FMT_TYPE_RAW]); + if (!vpu_fmt) + return -EINVAL; + + wave5_update_pix_fmt(&inst->dst_fmt, + VPU_FMT_TYPE_RAW, + initial_info->pic_width, + initial_info->pic_height, + vpu_fmt->v4l2_frmsize); } v4l2_event_queue_fh(fh, &vpu_event_src_ch); @@ -518,6 +451,8 @@ static void wave5_vpu_dec_finish_decode(struct vpu_instance *inst) if (q_status.report_queue_count == 0 && (q_status.instance_queue_count == 0 || dec_info.sequence_changed)) { dev_dbg(inst->dev->dev, "%s: finishing job.\n", __func__); + pm_runtime_mark_last_busy(inst->dev->dev); + pm_runtime_put_autosuspend(inst->dev->dev); v4l2_m2m_job_finish(inst->v4l2_m2m_dev, m2m_ctx); } } @@ -545,12 +480,12 @@ static int wave5_vpu_dec_enum_framesizes(struct file *f, void *fh, struct v4l2_f } fsize->type = V4L2_FRMSIZE_TYPE_CONTINUOUS; - fsize->stepwise.min_width = vpu_fmt->min_width; - fsize->stepwise.max_width = vpu_fmt->max_width; - fsize->stepwise.step_width = 1; - fsize->stepwise.min_height = vpu_fmt->min_height; - fsize->stepwise.max_height = vpu_fmt->max_height; - fsize->stepwise.step_height = 1; + fsize->stepwise.min_width = vpu_fmt->v4l2_frmsize->min_width; + fsize->stepwise.max_width = vpu_fmt->v4l2_frmsize->max_width; + fsize->stepwise.step_width = W5_DEC_CODEC_STEP_WIDTH; + fsize->stepwise.min_height = vpu_fmt->v4l2_frmsize->min_height; + fsize->stepwise.max_height = vpu_fmt->v4l2_frmsize->max_height; + fsize->stepwise.step_height = W5_DEC_CODEC_STEP_HEIGHT; return 0; } @@ -573,6 +508,7 @@ static int wave5_vpu_dec_try_fmt_cap(struct file *file, void *fh, struct v4l2_fo { struct vpu_instance *inst = wave5_to_vpu_inst(fh); struct dec_info *p_dec_info = &inst->codec_info->dec_info; + const struct v4l2_frmsize_stepwise *frmsize; const struct vpu_format *vpu_fmt; int width, height; @@ -586,14 +522,12 @@ static int wave5_vpu_dec_try_fmt_cap(struct file *file, void *fh, struct v4l2_fo width = inst->dst_fmt.width; height = inst->dst_fmt.height; f->fmt.pix_mp.pixelformat = inst->dst_fmt.pixelformat; - f->fmt.pix_mp.num_planes = inst->dst_fmt.num_planes; + frmsize = &dec_raw_frmsize; } else { - const struct v4l2_format_info *info = v4l2_format_info(vpu_fmt->v4l2_pix_fmt); - - width = clamp(f->fmt.pix_mp.width, vpu_fmt->min_width, vpu_fmt->max_width); - height = clamp(f->fmt.pix_mp.height, vpu_fmt->min_height, vpu_fmt->max_height); + width = f->fmt.pix_mp.width; + height = f->fmt.pix_mp.height; f->fmt.pix_mp.pixelformat = vpu_fmt->v4l2_pix_fmt; - f->fmt.pix_mp.num_planes = info->mem_planes; + frmsize = vpu_fmt->v4l2_frmsize; } if (p_dec_info->initial_info_obtained) { @@ -601,9 +535,8 @@ static int wave5_vpu_dec_try_fmt_cap(struct file *file, void *fh, struct v4l2_fo height = inst->dst_fmt.height; } - wave5_update_pix_fmt(&f->fmt.pix_mp, width, height); - f->fmt.pix_mp.flags = 0; - f->fmt.pix_mp.field = V4L2_FIELD_NONE; + wave5_update_pix_fmt(&f->fmt.pix_mp, VPU_FMT_TYPE_RAW, + width, height, frmsize); f->fmt.pix_mp.colorspace = inst->colorspace; f->fmt.pix_mp.ycbcr_enc = inst->ycbcr_enc; f->fmt.pix_mp.quantization = inst->quantization; @@ -715,7 +648,9 @@ static int wave5_vpu_dec_enum_fmt_out(struct file *file, void *fh, struct v4l2_f static int wave5_vpu_dec_try_fmt_out(struct file *file, void *fh, struct v4l2_format *f) { struct vpu_instance *inst = wave5_to_vpu_inst(fh); + const struct v4l2_frmsize_stepwise *frmsize; const struct vpu_format *vpu_fmt; + int width, height; dev_dbg(inst->dev->dev, "%s: fourcc: %u width: %u height: %u num_planes: %u colorspace: %u field: %u\n", @@ -724,20 +659,19 @@ static int wave5_vpu_dec_try_fmt_out(struct file *file, void *fh, struct v4l2_fo vpu_fmt = wave5_find_vpu_fmt(f->fmt.pix_mp.pixelformat, dec_fmt_list[VPU_FMT_TYPE_CODEC]); if (!vpu_fmt) { + width = inst->src_fmt.width; + height = inst->src_fmt.height; f->fmt.pix_mp.pixelformat = inst->src_fmt.pixelformat; - f->fmt.pix_mp.num_planes = inst->src_fmt.num_planes; - wave5_update_pix_fmt(&f->fmt.pix_mp, inst->src_fmt.width, inst->src_fmt.height); + frmsize = &dec_hevc_frmsize; } else { - int width = clamp(f->fmt.pix_mp.width, vpu_fmt->min_width, vpu_fmt->max_width); - int height = clamp(f->fmt.pix_mp.height, vpu_fmt->min_height, vpu_fmt->max_height); - + width = f->fmt.pix_mp.width; + height = f->fmt.pix_mp.height; f->fmt.pix_mp.pixelformat = vpu_fmt->v4l2_pix_fmt; - f->fmt.pix_mp.num_planes = 1; - wave5_update_pix_fmt(&f->fmt.pix_mp, width, height); + frmsize = vpu_fmt->v4l2_frmsize; } - f->fmt.pix_mp.flags = 0; - f->fmt.pix_mp.field = V4L2_FIELD_NONE; + wave5_update_pix_fmt(&f->fmt.pix_mp, VPU_FMT_TYPE_CODEC, + width, height, frmsize); return 0; } @@ -745,6 +679,7 @@ static int wave5_vpu_dec_try_fmt_out(struct file *file, void *fh, struct v4l2_fo static int wave5_vpu_dec_s_fmt_out(struct file *file, void *fh, struct v4l2_format *f) { struct vpu_instance *inst = wave5_to_vpu_inst(fh); + const struct vpu_format *vpu_fmt; int i, ret; dev_dbg(inst->dev->dev, @@ -779,7 +714,13 @@ static int wave5_vpu_dec_s_fmt_out(struct file *file, void *fh, struct v4l2_form inst->quantization = f->fmt.pix_mp.quantization; inst->xfer_func = f->fmt.pix_mp.xfer_func; - wave5_update_pix_fmt(&inst->dst_fmt, f->fmt.pix_mp.width, f->fmt.pix_mp.height); + vpu_fmt = wave5_find_vpu_fmt(inst->dst_fmt.pixelformat, dec_fmt_list[VPU_FMT_TYPE_RAW]); + if (!vpu_fmt) + return -EINVAL; + + wave5_update_pix_fmt(&inst->dst_fmt, VPU_FMT_TYPE_RAW, + f->fmt.pix_mp.width, f->fmt.pix_mp.height, + vpu_fmt->v4l2_frmsize); return 0; } @@ -1002,6 +943,7 @@ static int wave5_vpu_dec_queue_setup(struct vb2_queue *q, unsigned int *num_buff struct vpu_instance *inst = vb2_get_drv_priv(q); struct v4l2_pix_format_mplane inst_format = (q->type == V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE) ? inst->src_fmt : inst->dst_fmt; + unsigned int i; dev_dbg(inst->dev->dev, "%s: num_buffers: %u | num_planes: %u | type: %u\n", __func__, *num_buffers, *num_planes, q->type); @@ -1015,31 +957,9 @@ static int wave5_vpu_dec_queue_setup(struct vb2_queue *q, unsigned int *num_buff if (*num_buffers < inst->fbc_buf_count) *num_buffers = inst->fbc_buf_count; - if (*num_planes == 1) { - if (inst->output_format == FORMAT_422) - sizes[0] = inst_format.width * inst_format.height * 2; - else - sizes[0] = inst_format.width * inst_format.height * 3 / 2; - dev_dbg(inst->dev->dev, "%s: size[0]: %u\n", __func__, sizes[0]); - } else if (*num_planes == 2) { - sizes[0] = inst_format.width * inst_format.height; - if (inst->output_format == FORMAT_422) - sizes[1] = inst_format.width * inst_format.height; - else - sizes[1] = inst_format.width * inst_format.height / 2; - dev_dbg(inst->dev->dev, "%s: size[0]: %u | size[1]: %u\n", - __func__, sizes[0], sizes[1]); - } else if (*num_planes == 3) { - sizes[0] = inst_format.width * inst_format.height; - if (inst->output_format == FORMAT_422) { - sizes[1] = inst_format.width * inst_format.height / 2; - sizes[2] = inst_format.width * inst_format.height / 2; - } else { - sizes[1] = inst_format.width * inst_format.height / 4; - sizes[2] = inst_format.width * inst_format.height / 4; - } - dev_dbg(inst->dev->dev, "%s: size[0]: %u | size[1]: %u | size[2]: %u\n", - __func__, sizes[0], sizes[1], sizes[2]); + for (i = 0; i < *num_planes; i++) { + sizes[i] = inst_format.plane_fmt[i].sizeimage; + dev_dbg(inst->dev->dev, "%s: size[%u]: %u\n", __func__, i, sizes[i]); } } @@ -1055,6 +975,22 @@ static int wave5_prepare_fb(struct vpu_instance *inst) int ret, i; struct v4l2_m2m_buffer *buf, *n; struct v4l2_m2m_ctx *m2m_ctx = inst->v4l2_fh.m2m_ctx; + u32 bitdepth = inst->codec_info->dec_info.initial_info.luma_bitdepth; + + switch (bitdepth) { + case 8: + break; + case 10: + if (inst->std == W_HEVC_DEC && + inst->dev->attr.support_hevc10bit_dec) + break; + + fallthrough; + default: + dev_err(inst->dev->dev, "no support for %d bit depth\n", bitdepth); + + return -EINVAL; + } linear_num = v4l2_m2m_num_dst_bufs_ready(m2m_ctx); non_linear_num = inst->fbc_buf_count; @@ -1063,7 +999,7 @@ static int wave5_prepare_fb(struct vpu_instance *inst) struct frame_buffer *frame = &inst->frame_buf[i]; struct vpu_buf *vframe = &inst->frame_vbuf[i]; - fb_stride = inst->dst_fmt.width; + fb_stride = ALIGN(inst->dst_fmt.width * bitdepth / 8, 32); fb_height = ALIGN(inst->dst_fmt.height, 32); luma_size = fb_stride * fb_height; @@ -1382,6 +1318,7 @@ static int wave5_vpu_dec_start_streaming(struct vb2_queue *q, unsigned int count int ret = 0; dev_dbg(inst->dev->dev, "%s: type: %u\n", __func__, q->type); + pm_runtime_resume_and_get(inst->dev->dev); v4l2_m2m_update_start_streaming_state(m2m_ctx, q); @@ -1416,7 +1353,8 @@ static int wave5_vpu_dec_start_streaming(struct vb2_queue *q, unsigned int count if (ret) goto return_buffers; - if (inst->state == VPU_INST_STATE_INIT_SEQ) { + if (inst->state == VPU_INST_STATE_INIT_SEQ && + inst->dev->product_code == WAVE521C_CODE) { if (initial_info->luma_bitdepth != 8) { dev_info(inst->dev->dev, "%s: no support for %d bit depth", __func__, initial_info->luma_bitdepth); @@ -1424,14 +1362,17 @@ static int wave5_vpu_dec_start_streaming(struct vb2_queue *q, unsigned int count goto return_buffers; } } - } + } + pm_runtime_mark_last_busy(inst->dev->dev); + pm_runtime_put_autosuspend(inst->dev->dev); return ret; free_bitstream_vbuf: wave5_vdi_free_dma_memory(inst->dev, &inst->bitstream_vbuf); return_buffers: wave5_return_bufs(q, VB2_BUF_STATE_QUEUED); + pm_runtime_put_autosuspend(inst->dev->dev); return ret; } @@ -1442,6 +1383,16 @@ static int streamoff_output(struct vb2_queue *q) struct vb2_v4l2_buffer *buf; int ret; dma_addr_t new_rd_ptr; + struct dec_output_info dec_info; + unsigned int i; + + for (i = 0; i < v4l2_m2m_num_dst_bufs_ready(m2m_ctx); i++) { + ret = wave5_vpu_dec_set_disp_flag(inst, i); + if (ret) + dev_dbg(inst->dev->dev, + "%s: Setting display flag of buf index: %u, fail: %d\n", + __func__, i, ret); + } while ((buf = v4l2_m2m_src_buf_remove(m2m_ctx))) { dev_dbg(inst->dev->dev, "%s: (Multiplanar) buf type %4u | index %4u\n", @@ -1449,6 +1400,11 @@ static int streamoff_output(struct vb2_queue *q) v4l2_m2m_buf_done(buf, VB2_BUF_STATE_ERROR); } + while (wave5_vpu_dec_get_output_info(inst, &dec_info) == 0) { + if (dec_info.index_frame_display >= 0) + wave5_vpu_dec_set_disp_flag(inst, dec_info.index_frame_display); + } + ret = wave5_vpu_flush_instance(inst); if (ret) return ret; @@ -1517,6 +1473,7 @@ static void wave5_vpu_dec_stop_streaming(struct vb2_queue *q) bool check_cmd = TRUE; dev_dbg(inst->dev->dev, "%s: type: %u\n", __func__, q->type); + pm_runtime_resume_and_get(inst->dev->dev); while (check_cmd) { struct queue_status_info q_status; @@ -1531,7 +1488,7 @@ static void wave5_vpu_dec_stop_streaming(struct vb2_queue *q) break; if (wave5_vpu_dec_get_output_info(inst, &dec_output_info)) - dev_dbg(inst->dev->dev, "Getting decoding results from fw, fail\n"); + dev_dbg(inst->dev->dev, "there is no output info\n"); } v4l2_m2m_update_stop_streaming_state(m2m_ctx, q); @@ -1540,12 +1497,13 @@ static void wave5_vpu_dec_stop_streaming(struct vb2_queue *q) streamoff_output(q); else streamoff_capture(q); + + pm_runtime_mark_last_busy(inst->dev->dev); + pm_runtime_put_autosuspend(inst->dev->dev); } static const struct vb2_ops wave5_vpu_dec_vb2_ops = { .queue_setup = wave5_vpu_dec_queue_setup, - .wait_prepare = vb2_ops_wait_prepare, - .wait_finish = vb2_ops_wait_finish, .buf_queue = wave5_vpu_dec_buf_queue, .start_streaming = wave5_vpu_dec_start_streaming, .stop_streaming = wave5_vpu_dec_stop_streaming, @@ -1554,20 +1512,15 @@ static const struct vb2_ops wave5_vpu_dec_vb2_ops = { static void wave5_set_default_format(struct v4l2_pix_format_mplane *src_fmt, struct v4l2_pix_format_mplane *dst_fmt) { - unsigned int dst_pix_fmt = dec_fmt_list[VPU_FMT_TYPE_RAW][0].v4l2_pix_fmt; - const struct v4l2_format_info *dst_fmt_info = v4l2_format_info(dst_pix_fmt); - src_fmt->pixelformat = dec_fmt_list[VPU_FMT_TYPE_CODEC][0].v4l2_pix_fmt; - src_fmt->field = V4L2_FIELD_NONE; - src_fmt->flags = 0; - src_fmt->num_planes = 1; - wave5_update_pix_fmt(src_fmt, 720, 480); - - dst_fmt->pixelformat = dst_pix_fmt; - dst_fmt->field = V4L2_FIELD_NONE; - dst_fmt->flags = 0; - dst_fmt->num_planes = dst_fmt_info->mem_planes; - wave5_update_pix_fmt(dst_fmt, 736, 480); + wave5_update_pix_fmt(src_fmt, VPU_FMT_TYPE_CODEC, + W5_DEF_DEC_PIC_WIDTH, W5_DEF_DEC_PIC_HEIGHT, + &dec_hevc_frmsize); + + dst_fmt->pixelformat = dec_fmt_list[VPU_FMT_TYPE_RAW][0].v4l2_pix_fmt; + wave5_update_pix_fmt(dst_fmt, VPU_FMT_TYPE_RAW, + W5_DEF_DEC_PIC_WIDTH, W5_DEF_DEC_PIC_HEIGHT, + &dec_raw_frmsize); } static int wave5_vpu_dec_queue_init(void *priv, struct vb2_queue *src_vq, struct vb2_queue *dst_vq) @@ -1626,7 +1579,7 @@ static void wave5_vpu_dec_device_run(void *priv) int ret = 0; dev_dbg(inst->dev->dev, "%s: Fill the ring buffer with new bitstream data", __func__); - + pm_runtime_resume_and_get(inst->dev->dev); ret = fill_ringbuffer(inst); if (ret) { dev_warn(inst->dev->dev, "Filling ring buffer failed\n"); @@ -1709,6 +1662,8 @@ static void wave5_vpu_dec_device_run(void *priv) finish_job_and_return: dev_dbg(inst->dev->dev, "%s: leave and finish job", __func__); + pm_runtime_mark_last_busy(inst->dev->dev); + pm_runtime_put_autosuspend(inst->dev->dev); v4l2_m2m_job_finish(inst->v4l2_m2m_dev, m2m_ctx); } @@ -1810,7 +1765,6 @@ static int wave5_vpu_open_dec(struct file *filp) v4l2_fh_add(&inst->v4l2_fh); INIT_LIST_HEAD(&inst->list); - list_add_tail(&inst->list, &dev->instances); inst->v4l2_m2m_dev = inst->dev->v4l2_m2m_dec_dev; inst->v4l2_fh.m2m_ctx = @@ -1865,7 +1819,23 @@ static int wave5_vpu_open_dec(struct file *filp) goto cleanup_inst; } - wave5_vdi_allocate_sram(inst->dev); + /* + * For Wave515 SRAM memory was already allocated + * at wave5_vpu_dec_register_device() + */ + if (inst->dev->product_code != WAVE515_CODE) + wave5_vdi_allocate_sram(inst->dev); + + ret = mutex_lock_interruptible(&dev->dev_lock); + if (ret) + goto cleanup_inst; + + if (list_empty(&dev->instances)) + pm_runtime_use_autosuspend(inst->dev->dev); + + list_add_tail(&inst->list, &dev->instances); + + mutex_unlock(&dev->dev_lock); return 0; @@ -1893,6 +1863,13 @@ int wave5_vpu_dec_register_device(struct vpu_device *dev) struct video_device *vdev_dec; int ret; + /* + * Secondary AXI setup for Wave515 is done by INIT_VPU command, + * i.e. wave5_vpu_init(), that's why we allocate SRAM memory early. + */ + if (dev->product_code == WAVE515_CODE) + wave5_vdi_allocate_sram(dev); + vdev_dec = devm_kzalloc(dev->v4l2_dev.dev, sizeof(*vdev_dec), GFP_KERNEL); if (!vdev_dec) return -ENOMEM; @@ -1926,6 +1903,13 @@ int wave5_vpu_dec_register_device(struct vpu_device *dev) void wave5_vpu_dec_unregister_device(struct vpu_device *dev) { + /* + * Here is a freeing pair for Wave515 SRAM memory allocation + * happened at wave5_vpu_dec_register_device(). + */ + if (dev->product_code == WAVE515_CODE) + wave5_vdi_free_sram(dev); + video_unregister_device(dev->video_dev_dec); if (dev->v4l2_m2m_dec_dev) v4l2_m2m_release(dev->v4l2_m2m_dec_dev); diff --git a/drivers/media/platform/chips-media/wave5/wave5-vpu-enc.c b/drivers/media/platform/chips-media/wave5/wave5-vpu-enc.c index 8bbf9d10b467..1e5fc5f8b856 100644 --- a/drivers/media/platform/chips-media/wave5/wave5-vpu-enc.c +++ b/drivers/media/platform/chips-media/wave5/wave5-vpu-enc.c @@ -5,70 +5,90 @@ * Copyright (C) 2021-2023 CHIPS&MEDIA INC */ +#include <linux/pm_runtime.h> #include "wave5-helper.h" #define VPU_ENC_DEV_NAME "C&M Wave5 VPU encoder" #define VPU_ENC_DRV_NAME "wave5-enc" +static const struct v4l2_frmsize_stepwise enc_frmsize[FMT_TYPES] = { + [VPU_FMT_TYPE_CODEC] = { + .min_width = W5_MIN_ENC_PIC_WIDTH, + .max_width = W5_MAX_ENC_PIC_WIDTH, + .step_width = W5_ENC_CODEC_STEP_WIDTH, + .min_height = W5_MIN_ENC_PIC_HEIGHT, + .max_height = W5_MAX_ENC_PIC_HEIGHT, + .step_height = W5_ENC_CODEC_STEP_HEIGHT, + }, + [VPU_FMT_TYPE_RAW] = { + .min_width = W5_MIN_ENC_PIC_WIDTH, + .max_width = W5_MAX_ENC_PIC_WIDTH, + .step_width = W5_ENC_RAW_STEP_WIDTH, + .min_height = W5_MIN_ENC_PIC_HEIGHT, + .max_height = W5_MAX_ENC_PIC_HEIGHT, + .step_height = W5_ENC_RAW_STEP_HEIGHT, + }, +}; + static const struct vpu_format enc_fmt_list[FMT_TYPES][MAX_FMTS] = { [VPU_FMT_TYPE_CODEC] = { { .v4l2_pix_fmt = V4L2_PIX_FMT_HEVC, - .max_width = W5_MAX_ENC_PIC_WIDTH, - .min_width = W5_MIN_ENC_PIC_WIDTH, - .max_height = W5_MAX_ENC_PIC_HEIGHT, - .min_height = W5_MIN_ENC_PIC_HEIGHT, + .v4l2_frmsize = &enc_frmsize[VPU_FMT_TYPE_CODEC], }, { .v4l2_pix_fmt = V4L2_PIX_FMT_H264, - .max_width = W5_MAX_ENC_PIC_WIDTH, - .min_width = W5_MIN_ENC_PIC_WIDTH, - .max_height = W5_MAX_ENC_PIC_HEIGHT, - .min_height = W5_MIN_ENC_PIC_HEIGHT, + .v4l2_frmsize = &enc_frmsize[VPU_FMT_TYPE_CODEC], }, }, [VPU_FMT_TYPE_RAW] = { { .v4l2_pix_fmt = V4L2_PIX_FMT_YUV420, - .max_width = W5_MAX_ENC_PIC_WIDTH, - .min_width = W5_MIN_ENC_PIC_WIDTH, - .max_height = W5_MAX_ENC_PIC_HEIGHT, - .min_height = W5_MIN_ENC_PIC_HEIGHT, + .v4l2_frmsize = &enc_frmsize[VPU_FMT_TYPE_RAW], }, { .v4l2_pix_fmt = V4L2_PIX_FMT_NV12, - .max_width = W5_MAX_ENC_PIC_WIDTH, - .min_width = W5_MIN_ENC_PIC_WIDTH, - .max_height = W5_MAX_ENC_PIC_HEIGHT, - .min_height = W5_MIN_ENC_PIC_HEIGHT, + .v4l2_frmsize = &enc_frmsize[VPU_FMT_TYPE_RAW], }, { .v4l2_pix_fmt = V4L2_PIX_FMT_NV21, - .max_width = W5_MAX_ENC_PIC_WIDTH, - .min_width = W5_MIN_ENC_PIC_WIDTH, - .max_height = W5_MAX_ENC_PIC_HEIGHT, - .min_height = W5_MIN_ENC_PIC_HEIGHT, + .v4l2_frmsize = &enc_frmsize[VPU_FMT_TYPE_RAW], }, { .v4l2_pix_fmt = V4L2_PIX_FMT_YUV420M, - .max_width = W5_MAX_ENC_PIC_WIDTH, - .min_width = W5_MIN_ENC_PIC_WIDTH, - .max_height = W5_MAX_ENC_PIC_HEIGHT, - .min_height = W5_MIN_ENC_PIC_HEIGHT, + .v4l2_frmsize = &enc_frmsize[VPU_FMT_TYPE_RAW], }, { .v4l2_pix_fmt = V4L2_PIX_FMT_NV12M, - .max_width = W5_MAX_ENC_PIC_WIDTH, - .min_width = W5_MIN_ENC_PIC_WIDTH, - .max_height = W5_MAX_ENC_PIC_HEIGHT, - .min_height = W5_MIN_ENC_PIC_HEIGHT, + .v4l2_frmsize = &enc_frmsize[VPU_FMT_TYPE_RAW], }, { .v4l2_pix_fmt = V4L2_PIX_FMT_NV21M, - .max_width = W5_MAX_ENC_PIC_WIDTH, - .min_width = W5_MIN_ENC_PIC_WIDTH, - .max_height = W5_MAX_ENC_PIC_HEIGHT, - .min_height = W5_MIN_ENC_PIC_HEIGHT, + .v4l2_frmsize = &enc_frmsize[VPU_FMT_TYPE_RAW], + }, + { + .v4l2_pix_fmt = V4L2_PIX_FMT_YUV422P, + .v4l2_frmsize = &enc_frmsize[VPU_FMT_TYPE_RAW], + }, + { + .v4l2_pix_fmt = V4L2_PIX_FMT_NV16, + .v4l2_frmsize = &enc_frmsize[VPU_FMT_TYPE_RAW], + }, + { + .v4l2_pix_fmt = V4L2_PIX_FMT_NV61, + .v4l2_frmsize = &enc_frmsize[VPU_FMT_TYPE_RAW], + }, + { + .v4l2_pix_fmt = V4L2_PIX_FMT_YUV422M, + .v4l2_frmsize = &enc_frmsize[VPU_FMT_TYPE_RAW], + }, + { + .v4l2_pix_fmt = V4L2_PIX_FMT_NV16M, + .v4l2_frmsize = &enc_frmsize[VPU_FMT_TYPE_RAW], + }, + { + .v4l2_pix_fmt = V4L2_PIX_FMT_NV61M, + .v4l2_frmsize = &enc_frmsize[VPU_FMT_TYPE_RAW], }, } }; @@ -105,46 +125,6 @@ invalid_state_switch: return -EINVAL; } -static void wave5_update_pix_fmt(struct v4l2_pix_format_mplane *pix_mp, unsigned int width, - unsigned int height) -{ - switch (pix_mp->pixelformat) { - case V4L2_PIX_FMT_YUV420: - case V4L2_PIX_FMT_NV12: - case V4L2_PIX_FMT_NV21: - pix_mp->width = width; - pix_mp->height = height; - pix_mp->plane_fmt[0].bytesperline = round_up(width, 32); - pix_mp->plane_fmt[0].sizeimage = round_up(width, 32) * height * 3 / 2; - break; - case V4L2_PIX_FMT_YUV420M: - pix_mp->width = width; - pix_mp->height = height; - pix_mp->plane_fmt[0].bytesperline = round_up(width, 32); - pix_mp->plane_fmt[0].sizeimage = round_up(width, 32) * height; - pix_mp->plane_fmt[1].bytesperline = round_up(width, 32) / 2; - pix_mp->plane_fmt[1].sizeimage = round_up(width, 32) * height / 4; - pix_mp->plane_fmt[2].bytesperline = round_up(width, 32) / 2; - pix_mp->plane_fmt[2].sizeimage = round_up(width, 32) * height / 4; - break; - case V4L2_PIX_FMT_NV12M: - case V4L2_PIX_FMT_NV21M: - pix_mp->width = width; - pix_mp->height = height; - pix_mp->plane_fmt[0].bytesperline = round_up(width, 32); - pix_mp->plane_fmt[0].sizeimage = round_up(width, 32) * height; - pix_mp->plane_fmt[1].bytesperline = round_up(width, 32); - pix_mp->plane_fmt[1].sizeimage = round_up(width, 32) * height / 2; - break; - default: - pix_mp->width = width; - pix_mp->height = height; - pix_mp->plane_fmt[0].bytesperline = 0; - pix_mp->plane_fmt[0].sizeimage = width * height / 8 * 3; - break; - } -} - static int start_encode(struct vpu_instance *inst, u32 *fail_res) { struct v4l2_m2m_ctx *m2m_ctx = inst->v4l2_fh.m2m_ctx; @@ -153,13 +133,26 @@ static int start_encode(struct vpu_instance *inst, u32 *fail_res) struct vb2_v4l2_buffer *dst_buf; struct frame_buffer frame_buf; struct enc_param pic_param; - u32 stride = ALIGN(inst->dst_fmt.width, 32); - u32 luma_size = (stride * inst->dst_fmt.height); - u32 chroma_size = ((stride / 2) * (inst->dst_fmt.height / 2)); + const struct v4l2_format_info *info; + u32 stride = inst->src_fmt.plane_fmt[0].bytesperline; + u32 luma_size = 0; + u32 chroma_size = 0; memset(&pic_param, 0, sizeof(struct enc_param)); memset(&frame_buf, 0, sizeof(struct frame_buffer)); + info = v4l2_format_info(inst->src_fmt.pixelformat); + if (!info) + return -EINVAL; + + if (info->mem_planes == 1) { + luma_size = stride * inst->dst_fmt.height; + chroma_size = luma_size / (info->hdiv * info->vdiv); + } else { + luma_size = inst->src_fmt.plane_fmt[0].sizeimage; + chroma_size = inst->src_fmt.plane_fmt[1].sizeimage; + } + dst_buf = v4l2_m2m_next_dst_buf(m2m_ctx); if (!dst_buf) { dev_dbg(inst->dev->dev, "%s: No destination buffer found\n", __func__); @@ -359,13 +352,8 @@ static int wave5_vpu_enc_enum_framesizes(struct file *f, void *fh, struct v4l2_f return -EINVAL; } - fsize->type = V4L2_FRMSIZE_TYPE_CONTINUOUS; - fsize->stepwise.min_width = vpu_fmt->min_width; - fsize->stepwise.max_width = vpu_fmt->max_width; - fsize->stepwise.step_width = 1; - fsize->stepwise.min_height = vpu_fmt->min_height; - fsize->stepwise.max_height = vpu_fmt->max_height; - fsize->stepwise.step_height = 1; + fsize->type = V4L2_FRMSIZE_TYPE_STEPWISE; + fsize->stepwise = enc_frmsize[VPU_FMT_TYPE_CODEC]; return 0; } @@ -390,7 +378,9 @@ static int wave5_vpu_enc_enum_fmt_cap(struct file *file, void *fh, struct v4l2_f static int wave5_vpu_enc_try_fmt_cap(struct file *file, void *fh, struct v4l2_format *f) { struct vpu_instance *inst = wave5_to_vpu_inst(fh); + const struct v4l2_frmsize_stepwise *frmsize; const struct vpu_format *vpu_fmt; + int width, height; dev_dbg(inst->dev->dev, "%s: fourcc: %u width: %u height: %u num_planes: %u field: %u\n", __func__, f->fmt.pix_mp.pixelformat, f->fmt.pix_mp.width, f->fmt.pix_mp.height, @@ -398,20 +388,19 @@ static int wave5_vpu_enc_try_fmt_cap(struct file *file, void *fh, struct v4l2_fo vpu_fmt = wave5_find_vpu_fmt(f->fmt.pix_mp.pixelformat, enc_fmt_list[VPU_FMT_TYPE_CODEC]); if (!vpu_fmt) { + width = inst->dst_fmt.width; + height = inst->dst_fmt.height; f->fmt.pix_mp.pixelformat = inst->dst_fmt.pixelformat; - f->fmt.pix_mp.num_planes = inst->dst_fmt.num_planes; - wave5_update_pix_fmt(&f->fmt.pix_mp, inst->dst_fmt.width, inst->dst_fmt.height); + frmsize = &enc_frmsize[VPU_FMT_TYPE_CODEC]; } else { - int width = clamp(f->fmt.pix_mp.width, vpu_fmt->min_width, vpu_fmt->max_width); - int height = clamp(f->fmt.pix_mp.height, vpu_fmt->min_height, vpu_fmt->max_height); - + width = f->fmt.pix_mp.width; + height = f->fmt.pix_mp.height; f->fmt.pix_mp.pixelformat = vpu_fmt->v4l2_pix_fmt; - f->fmt.pix_mp.num_planes = 1; - wave5_update_pix_fmt(&f->fmt.pix_mp, width, height); + frmsize = vpu_fmt->v4l2_frmsize; } - f->fmt.pix_mp.flags = 0; - f->fmt.pix_mp.field = V4L2_FIELD_NONE; + wave5_update_pix_fmt(&f->fmt.pix_mp, VPU_FMT_TYPE_CODEC, + width, height, frmsize); f->fmt.pix_mp.colorspace = inst->colorspace; f->fmt.pix_mp.ycbcr_enc = inst->ycbcr_enc; f->fmt.pix_mp.quantization = inst->quantization; @@ -498,7 +487,9 @@ static int wave5_vpu_enc_enum_fmt_out(struct file *file, void *fh, struct v4l2_f static int wave5_vpu_enc_try_fmt_out(struct file *file, void *fh, struct v4l2_format *f) { struct vpu_instance *inst = wave5_to_vpu_inst(fh); + const struct v4l2_frmsize_stepwise *frmsize; const struct vpu_format *vpu_fmt; + int width, height; dev_dbg(inst->dev->dev, "%s: fourcc: %u width: %u height: %u num_planes: %u field: %u\n", __func__, f->fmt.pix_mp.pixelformat, f->fmt.pix_mp.width, f->fmt.pix_mp.height, @@ -506,28 +497,27 @@ static int wave5_vpu_enc_try_fmt_out(struct file *file, void *fh, struct v4l2_fo vpu_fmt = wave5_find_vpu_fmt(f->fmt.pix_mp.pixelformat, enc_fmt_list[VPU_FMT_TYPE_RAW]); if (!vpu_fmt) { + width = inst->src_fmt.width; + height = inst->src_fmt.height; f->fmt.pix_mp.pixelformat = inst->src_fmt.pixelformat; - f->fmt.pix_mp.num_planes = inst->src_fmt.num_planes; - wave5_update_pix_fmt(&f->fmt.pix_mp, inst->src_fmt.width, inst->src_fmt.height); + frmsize = &enc_frmsize[VPU_FMT_TYPE_RAW]; } else { - int width = clamp(f->fmt.pix_mp.width, vpu_fmt->min_width, vpu_fmt->max_width); - int height = clamp(f->fmt.pix_mp.height, vpu_fmt->min_height, vpu_fmt->max_height); - const struct v4l2_format_info *info = v4l2_format_info(vpu_fmt->v4l2_pix_fmt); - + width = f->fmt.pix_mp.width; + height = f->fmt.pix_mp.height; f->fmt.pix_mp.pixelformat = vpu_fmt->v4l2_pix_fmt; - f->fmt.pix_mp.num_planes = info->mem_planes; - wave5_update_pix_fmt(&f->fmt.pix_mp, width, height); + frmsize = vpu_fmt->v4l2_frmsize; } - f->fmt.pix_mp.flags = 0; - f->fmt.pix_mp.field = V4L2_FIELD_NONE; - + wave5_update_pix_fmt(&f->fmt.pix_mp, VPU_FMT_TYPE_RAW, + width, height, frmsize); return 0; } static int wave5_vpu_enc_s_fmt_out(struct file *file, void *fh, struct v4l2_format *f) { struct vpu_instance *inst = wave5_to_vpu_inst(fh); + const struct vpu_format *vpu_fmt; + const struct v4l2_format_info *info; int i, ret; dev_dbg(inst->dev->dev, "%s: fourcc: %u width: %u height: %u num_planes: %u field: %u\n", @@ -549,16 +539,20 @@ static int wave5_vpu_enc_s_fmt_out(struct file *file, void *fh, struct v4l2_form inst->src_fmt.plane_fmt[i].sizeimage = f->fmt.pix_mp.plane_fmt[i].sizeimage; } - if (inst->src_fmt.pixelformat == V4L2_PIX_FMT_NV12 || - inst->src_fmt.pixelformat == V4L2_PIX_FMT_NV12M) { - inst->cbcr_interleave = true; - inst->nv21 = false; - } else if (inst->src_fmt.pixelformat == V4L2_PIX_FMT_NV21 || - inst->src_fmt.pixelformat == V4L2_PIX_FMT_NV21M) { - inst->cbcr_interleave = true; + info = v4l2_format_info(inst->src_fmt.pixelformat); + if (!info) + return -EINVAL; + + inst->cbcr_interleave = (info->comp_planes == 2) ? true : false; + + switch (inst->src_fmt.pixelformat) { + case V4L2_PIX_FMT_NV21: + case V4L2_PIX_FMT_NV21M: + case V4L2_PIX_FMT_NV61: + case V4L2_PIX_FMT_NV61M: inst->nv21 = true; - } else { - inst->cbcr_interleave = false; + break; + default: inst->nv21 = false; } @@ -567,7 +561,15 @@ static int wave5_vpu_enc_s_fmt_out(struct file *file, void *fh, struct v4l2_form inst->quantization = f->fmt.pix_mp.quantization; inst->xfer_func = f->fmt.pix_mp.xfer_func; - wave5_update_pix_fmt(&inst->dst_fmt, f->fmt.pix_mp.width, f->fmt.pix_mp.height); + vpu_fmt = wave5_find_vpu_fmt(inst->dst_fmt.pixelformat, enc_fmt_list[VPU_FMT_TYPE_CODEC]); + if (!vpu_fmt) + return -EINVAL; + + wave5_update_pix_fmt(&inst->dst_fmt, VPU_FMT_TYPE_CODEC, + f->fmt.pix_mp.width, f->fmt.pix_mp.height, + vpu_fmt->v4l2_frmsize); + inst->conf_win.width = inst->dst_fmt.width; + inst->conf_win.height = inst->dst_fmt.height; return 0; } @@ -583,12 +585,17 @@ static int wave5_vpu_enc_g_selection(struct file *file, void *fh, struct v4l2_se switch (s->target) { case V4L2_SEL_TGT_CROP_DEFAULT: case V4L2_SEL_TGT_CROP_BOUNDS: - case V4L2_SEL_TGT_CROP: s->r.left = 0; s->r.top = 0; s->r.width = inst->dst_fmt.width; s->r.height = inst->dst_fmt.height; break; + case V4L2_SEL_TGT_CROP: + s->r.left = 0; + s->r.top = 0; + s->r.width = inst->conf_win.width; + s->r.height = inst->conf_win.height; + break; default: return -EINVAL; } @@ -611,8 +618,10 @@ static int wave5_vpu_enc_s_selection(struct file *file, void *fh, struct v4l2_se s->r.left = 0; s->r.top = 0; - s->r.width = inst->src_fmt.width; - s->r.height = inst->src_fmt.height; + s->r.width = min(s->r.width, inst->dst_fmt.width); + s->r.height = min(s->r.height, inst->dst_fmt.height); + + inst->conf_win = s->r; return 0; } @@ -1061,6 +1070,9 @@ static int wave5_vpu_enc_s_ctrl(struct v4l2_ctrl *ctrl) case V4L2_CID_MPEG_VIDEO_H264_ENTROPY_MODE: inst->enc_param.entropy_coding_mode = ctrl->val; break; + case V4L2_CID_MPEG_VIDEO_PREPEND_SPSPPS_TO_IDR: + inst->enc_param.forced_idr_header_enable = ctrl->val; + break; case V4L2_CID_MIN_BUFFERS_FOR_OUTPUT: break; default: @@ -1125,13 +1137,23 @@ static void wave5_vpu_enc_buf_queue(struct vb2_buffer *vb) v4l2_m2m_buf_queue(m2m_ctx, vbuf); } -static void wave5_set_enc_openparam(struct enc_open_param *open_param, - struct vpu_instance *inst) +static int wave5_set_enc_openparam(struct enc_open_param *open_param, + struct vpu_instance *inst) { struct enc_wave_param input = inst->enc_param; + const struct v4l2_format_info *info; u32 num_ctu_row = ALIGN(inst->dst_fmt.height, 64) / 64; u32 num_mb_row = ALIGN(inst->dst_fmt.height, 16) / 16; + info = v4l2_format_info(inst->src_fmt.pixelformat); + if (!info) + return -EINVAL; + + if (info->hdiv == 2 && info->vdiv == 1) + open_param->src_format = FORMAT_422; + else + open_param->src_format = FORMAT_420; + open_param->wave_param.gop_preset_idx = PRESET_IDX_IPP_SINGLE; open_param->wave_param.hvs_qp_scale = 2; open_param->wave_param.hvs_max_delta_qp = 10; @@ -1147,8 +1169,8 @@ static void wave5_set_enc_openparam(struct enc_open_param *open_param, open_param->wave_param.lambda_scaling_enable = 1; open_param->line_buf_int_en = true; - open_param->pic_width = inst->dst_fmt.width; - open_param->pic_height = inst->dst_fmt.height; + open_param->pic_width = inst->conf_win.width; + open_param->pic_height = inst->conf_win.height; open_param->frame_rate_info = inst->frame_rate; open_param->rc_enable = inst->rc_enable; if (inst->rc_enable) { @@ -1219,6 +1241,9 @@ static void wave5_set_enc_openparam(struct enc_open_param *open_param, else open_param->wave_param.intra_refresh_arg = num_ctu_row; } + open_param->wave_param.forced_idr_header_enable = input.forced_idr_header_enable; + + return 0; } static int initialize_sequence(struct vpu_instance *inst) @@ -1247,7 +1272,7 @@ static int initialize_sequence(struct vpu_instance *inst) __func__, initial_info.min_frame_buffer_count, initial_info.min_src_frame_count); inst->min_src_buf_count = initial_info.min_src_frame_count + - COMMAND_QUEUE_DEPTH; + WAVE521_COMMAND_QUEUE_DEPTH; ctrl = v4l2_ctrl_find(&inst->v4l2_ctrl_hdl, V4L2_CID_MIN_BUFFERS_FOR_OUTPUT); @@ -1306,6 +1331,7 @@ static int wave5_vpu_enc_start_streaming(struct vb2_queue *q, unsigned int count struct v4l2_m2m_ctx *m2m_ctx = inst->v4l2_fh.m2m_ctx; int ret = 0; + pm_runtime_resume_and_get(inst->dev->dev); v4l2_m2m_update_start_streaming_state(m2m_ctx, q); if (inst->state == VPU_INST_STATE_NONE && q->type == V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE) { @@ -1313,7 +1339,12 @@ static int wave5_vpu_enc_start_streaming(struct vb2_queue *q, unsigned int count memset(&open_param, 0, sizeof(struct enc_open_param)); - wave5_set_enc_openparam(&open_param, inst); + ret = wave5_set_enc_openparam(&open_param, inst); + if (ret) { + dev_dbg(inst->dev->dev, "%s: wave5_set_enc_openparam, fail: %d\n", + __func__, ret); + goto return_buffers; + } ret = wave5_vpu_enc_open(inst, &open_param); if (ret) { @@ -1360,9 +1391,13 @@ static int wave5_vpu_enc_start_streaming(struct vb2_queue *q, unsigned int count if (ret) goto return_buffers; + pm_runtime_mark_last_busy(inst->dev->dev); + pm_runtime_put_autosuspend(inst->dev->dev); return 0; return_buffers: wave5_return_bufs(q, VB2_BUF_STATE_QUEUED); + pm_runtime_mark_last_busy(inst->dev->dev); + pm_runtime_put_autosuspend(inst->dev->dev); return ret; } @@ -1404,6 +1439,7 @@ static void wave5_vpu_enc_stop_streaming(struct vb2_queue *q) */ dev_dbg(inst->dev->dev, "%s: type: %u\n", __func__, q->type); + pm_runtime_resume_and_get(inst->dev->dev); if (wave5_vpu_both_queues_are_streaming(inst)) switch_state(inst, VPU_INST_STATE_STOP); @@ -1428,12 +1464,13 @@ static void wave5_vpu_enc_stop_streaming(struct vb2_queue *q) streamoff_output(inst, q); else streamoff_capture(inst, q); + + pm_runtime_mark_last_busy(inst->dev->dev); + pm_runtime_put_autosuspend(inst->dev->dev); } static const struct vb2_ops wave5_vpu_enc_vb2_ops = { .queue_setup = wave5_vpu_enc_queue_setup, - .wait_prepare = vb2_ops_wait_prepare, - .wait_finish = vb2_ops_wait_finish, .buf_queue = wave5_vpu_enc_buf_queue, .start_streaming = wave5_vpu_enc_start_streaming, .stop_streaming = wave5_vpu_enc_stop_streaming, @@ -1442,20 +1479,15 @@ static const struct vb2_ops wave5_vpu_enc_vb2_ops = { static void wave5_set_default_format(struct v4l2_pix_format_mplane *src_fmt, struct v4l2_pix_format_mplane *dst_fmt) { - unsigned int src_pix_fmt = enc_fmt_list[VPU_FMT_TYPE_RAW][0].v4l2_pix_fmt; - const struct v4l2_format_info *src_fmt_info = v4l2_format_info(src_pix_fmt); - - src_fmt->pixelformat = src_pix_fmt; - src_fmt->field = V4L2_FIELD_NONE; - src_fmt->flags = 0; - src_fmt->num_planes = src_fmt_info->mem_planes; - wave5_update_pix_fmt(src_fmt, 416, 240); + src_fmt->pixelformat = enc_fmt_list[VPU_FMT_TYPE_RAW][0].v4l2_pix_fmt; + wave5_update_pix_fmt(src_fmt, VPU_FMT_TYPE_RAW, + W5_DEF_ENC_PIC_WIDTH, W5_DEF_ENC_PIC_HEIGHT, + &enc_frmsize[VPU_FMT_TYPE_RAW]); dst_fmt->pixelformat = enc_fmt_list[VPU_FMT_TYPE_CODEC][0].v4l2_pix_fmt; - dst_fmt->field = V4L2_FIELD_NONE; - dst_fmt->flags = 0; - dst_fmt->num_planes = 1; - wave5_update_pix_fmt(dst_fmt, 416, 240); + wave5_update_pix_fmt(dst_fmt, VPU_FMT_TYPE_CODEC, + W5_DEF_ENC_PIC_WIDTH, W5_DEF_ENC_PIC_HEIGHT, + &enc_frmsize[VPU_FMT_TYPE_CODEC]); } static int wave5_vpu_enc_queue_init(void *priv, struct vb2_queue *src_vq, struct vb2_queue *dst_vq) @@ -1474,6 +1506,7 @@ static void wave5_vpu_enc_device_run(void *priv) u32 fail_res = 0; int ret = 0; + pm_runtime_resume_and_get(inst->dev->dev); switch (inst->state) { case VPU_INST_STATE_PIC_RUN: ret = start_encode(inst, &fail_res); @@ -1487,6 +1520,8 @@ static void wave5_vpu_enc_device_run(void *priv) break; } dev_dbg(inst->dev->dev, "%s: leave with active job", __func__); + pm_runtime_mark_last_busy(inst->dev->dev); + pm_runtime_put_autosuspend(inst->dev->dev); return; default: WARN(1, "Execution of a job in state %s is invalid.\n", @@ -1494,6 +1529,8 @@ static void wave5_vpu_enc_device_run(void *priv) break; } dev_dbg(inst->dev->dev, "%s: leave and finish job", __func__); + pm_runtime_mark_last_busy(inst->dev->dev); + pm_runtime_put_autosuspend(inst->dev->dev); v4l2_m2m_job_finish(inst->v4l2_m2m_dev, m2m_ctx); } @@ -1554,7 +1591,6 @@ static int wave5_vpu_open_enc(struct file *filp) v4l2_fh_add(&inst->v4l2_fh); INIT_LIST_HEAD(&inst->list); - list_add_tail(&inst->list, &dev->instances); inst->v4l2_m2m_dev = inst->dev->v4l2_m2m_enc_dev; inst->v4l2_fh.m2m_ctx = @@ -1702,6 +1738,9 @@ static int wave5_vpu_open_enc(struct file *filp) 0, 1, 1, 0); v4l2_ctrl_new_std(v4l2_ctrl_hdl, &wave5_vpu_enc_ctrl_ops, V4L2_CID_MIN_BUFFERS_FOR_OUTPUT, 1, 32, 1, 1); + v4l2_ctrl_new_std(v4l2_ctrl_hdl, &wave5_vpu_enc_ctrl_ops, + V4L2_CID_MPEG_VIDEO_PREPEND_SPSPPS_TO_IDR, + 0, 1, 1, 0); if (v4l2_ctrl_hdl->error) { ret = -ENODEV; @@ -1712,6 +1751,8 @@ static int wave5_vpu_open_enc(struct file *filp) v4l2_ctrl_handler_setup(v4l2_ctrl_hdl); wave5_set_default_format(&inst->src_fmt, &inst->dst_fmt); + inst->conf_win.width = inst->dst_fmt.width; + inst->conf_win.height = inst->dst_fmt.height; inst->colorspace = V4L2_COLORSPACE_REC709; inst->ycbcr_enc = V4L2_YCBCR_ENC_DEFAULT; inst->quantization = V4L2_QUANTIZATION_DEFAULT; @@ -1729,6 +1770,17 @@ static int wave5_vpu_open_enc(struct file *filp) wave5_vdi_allocate_sram(inst->dev); + ret = mutex_lock_interruptible(&dev->dev_lock); + if (ret) + goto cleanup_inst; + + if (list_empty(&dev->instances)) + pm_runtime_use_autosuspend(inst->dev->dev); + + list_add_tail(&inst->list, &dev->instances); + + mutex_unlock(&dev->dev_lock); + return 0; cleanup_inst: diff --git a/drivers/media/platform/chips-media/wave5/wave5-vpu.c b/drivers/media/platform/chips-media/wave5/wave5-vpu.c index 1b3df5b04249..e1715d3f43b0 100644 --- a/drivers/media/platform/chips-media/wave5/wave5-vpu.c +++ b/drivers/media/platform/chips-media/wave5/wave5-vpu.c @@ -10,6 +10,8 @@ #include <linux/clk.h> #include <linux/firmware.h> #include <linux/interrupt.h> +#include <linux/pm_runtime.h> +#include <linux/reset.h> #include "wave5-vpu.h" #include "wave5-regdefine.h" #include "wave5-vpuconfig.h" @@ -24,8 +26,12 @@ struct wave5_match_data { int flags; const char *fw_name; + u32 sram_size; }; +static int vpu_poll_interval = 5; +module_param(vpu_poll_interval, int, 0644); + int wave5_vpu_wait_interrupt(struct vpu_instance *inst, unsigned int timeout) { int ret; @@ -40,7 +46,7 @@ int wave5_vpu_wait_interrupt(struct vpu_instance *inst, unsigned int timeout) return 0; } -static irqreturn_t wave5_vpu_irq_thread(int irq, void *dev_id) +static void wave5_vpu_handle_irq(void *dev_id) { u32 seq_done; u32 cmd_done; @@ -48,42 +54,73 @@ static irqreturn_t wave5_vpu_irq_thread(int irq, void *dev_id) struct vpu_instance *inst; struct vpu_device *dev = dev_id; - if (wave5_vdi_read_register(dev, W5_VPU_VPU_INT_STS)) { - irq_reason = wave5_vdi_read_register(dev, W5_VPU_VINT_REASON); - wave5_vdi_write_register(dev, W5_VPU_VINT_REASON_CLR, irq_reason); - wave5_vdi_write_register(dev, W5_VPU_VINT_CLEAR, 0x1); - - list_for_each_entry(inst, &dev->instances, list) { - seq_done = wave5_vdi_read_register(dev, W5_RET_SEQ_DONE_INSTANCE_INFO); - cmd_done = wave5_vdi_read_register(dev, W5_RET_QUEUE_CMD_DONE_INST); - - if (irq_reason & BIT(INT_WAVE5_INIT_SEQ) || - irq_reason & BIT(INT_WAVE5_ENC_SET_PARAM)) { - if (seq_done & BIT(inst->id)) { - seq_done &= ~BIT(inst->id); - wave5_vdi_write_register(dev, W5_RET_SEQ_DONE_INSTANCE_INFO, - seq_done); - complete(&inst->irq_done); - } + irq_reason = wave5_vdi_read_register(dev, W5_VPU_VINT_REASON); + seq_done = wave5_vdi_read_register(dev, W5_RET_SEQ_DONE_INSTANCE_INFO); + cmd_done = wave5_vdi_read_register(dev, W5_RET_QUEUE_CMD_DONE_INST); + wave5_vdi_write_register(dev, W5_VPU_VINT_REASON_CLR, irq_reason); + wave5_vdi_write_register(dev, W5_VPU_VINT_CLEAR, 0x1); + + list_for_each_entry(inst, &dev->instances, list) { + + if (irq_reason & BIT(INT_WAVE5_INIT_SEQ) || + irq_reason & BIT(INT_WAVE5_ENC_SET_PARAM)) { + if (dev->product_code == WAVE515_CODE && + (cmd_done & BIT(inst->id))) { + cmd_done &= ~BIT(inst->id); + wave5_vdi_write_register(dev, W5_RET_QUEUE_CMD_DONE_INST, + cmd_done); + complete(&inst->irq_done); + } else if (seq_done & BIT(inst->id)) { + seq_done &= ~BIT(inst->id); + wave5_vdi_write_register(dev, W5_RET_SEQ_DONE_INSTANCE_INFO, + seq_done); + complete(&inst->irq_done); } + } - if (irq_reason & BIT(INT_WAVE5_DEC_PIC) || - irq_reason & BIT(INT_WAVE5_ENC_PIC)) { - if (cmd_done & BIT(inst->id)) { - cmd_done &= ~BIT(inst->id); - wave5_vdi_write_register(dev, W5_RET_QUEUE_CMD_DONE_INST, - cmd_done); - inst->ops->finish_process(inst); - } + if (irq_reason & BIT(INT_WAVE5_DEC_PIC) || + irq_reason & BIT(INT_WAVE5_ENC_PIC)) { + if (cmd_done & BIT(inst->id)) { + cmd_done &= ~BIT(inst->id); + wave5_vdi_write_register(dev, W5_RET_QUEUE_CMD_DONE_INST, + cmd_done); + inst->ops->finish_process(inst); } - - wave5_vpu_clear_interrupt(inst, irq_reason); } + + wave5_vpu_clear_interrupt(inst, irq_reason); } +} + +static irqreturn_t wave5_vpu_irq_thread(int irq, void *dev_id) +{ + struct vpu_device *dev = dev_id; + + if (wave5_vdi_read_register(dev, W5_VPU_VPU_INT_STS)) + wave5_vpu_handle_irq(dev); return IRQ_HANDLED; } +static void wave5_vpu_irq_work_fn(struct kthread_work *work) +{ + struct vpu_device *dev = container_of(work, struct vpu_device, work); + + if (wave5_vdi_read_register(dev, W5_VPU_VPU_INT_STS)) + wave5_vpu_handle_irq(dev); +} + +static enum hrtimer_restart wave5_vpu_timer_callback(struct hrtimer *timer) +{ + struct vpu_device *dev = + container_of(timer, struct vpu_device, hrtimer); + + kthread_queue_work(dev->worker, &dev->work); + hrtimer_forward_now(timer, ns_to_ktime(vpu_poll_interval * NSEC_PER_MSEC)); + + return HRTIMER_RESTART; +} + static int wave5_vpu_load_firmware(struct device *dev, const char *fw_name, u32 *revision) { @@ -117,6 +154,45 @@ static int wave5_vpu_load_firmware(struct device *dev, const char *fw_name, return 0; } +static __maybe_unused int wave5_pm_suspend(struct device *dev) +{ + struct vpu_device *vpu = dev_get_drvdata(dev); + + if (pm_runtime_suspended(dev)) + return 0; + + if (vpu->irq < 0) + hrtimer_cancel(&vpu->hrtimer); + + wave5_vpu_sleep_wake(dev, true, NULL, 0); + clk_bulk_disable_unprepare(vpu->num_clks, vpu->clks); + + return 0; +} + +static __maybe_unused int wave5_pm_resume(struct device *dev) +{ + struct vpu_device *vpu = dev_get_drvdata(dev); + int ret = 0; + + wave5_vpu_sleep_wake(dev, false, NULL, 0); + ret = clk_bulk_prepare_enable(vpu->num_clks, vpu->clks); + if (ret) { + dev_err(dev, "Enabling clocks, fail: %d\n", ret); + return ret; + } + + if (vpu->irq < 0 && !hrtimer_active(&vpu->hrtimer)) + hrtimer_start(&vpu->hrtimer, ns_to_ktime(vpu->vpu_poll_interval * NSEC_PER_MSEC), + HRTIMER_MODE_REL_PINNED); + + return ret; +} + +static const struct dev_pm_ops wave5_pm_ops = { + SET_RUNTIME_PM_OPS(wave5_pm_suspend, wave5_pm_resume, NULL) +}; + static int wave5_vpu_probe(struct platform_device *pdev) { int ret; @@ -151,6 +227,16 @@ static int wave5_vpu_probe(struct platform_device *pdev) dev_set_drvdata(&pdev->dev, dev); dev->dev = &pdev->dev; + dev->resets = devm_reset_control_array_get_optional_exclusive(&pdev->dev); + if (IS_ERR(dev->resets)) { + return dev_err_probe(&pdev->dev, PTR_ERR(dev->resets), + "Failed to get reset control\n"); + } + + ret = reset_control_deassert(dev->resets); + if (ret) + return dev_err_probe(&pdev->dev, ret, "Failed to deassert resets\n"); + ret = devm_clk_bulk_get_all(&pdev->dev, &dev->clks); /* continue without clock, assume externally managed */ @@ -163,20 +249,15 @@ static int wave5_vpu_probe(struct platform_device *pdev) ret = clk_bulk_prepare_enable(dev->num_clks, dev->clks); if (ret) { dev_err(&pdev->dev, "Enabling clocks, fail: %d\n", ret); - return ret; - } - - ret = of_property_read_u32(pdev->dev.of_node, "sram-size", - &dev->sram_size); - if (ret) { - dev_warn(&pdev->dev, "sram-size not found\n"); - dev->sram_size = 0; + goto err_reset_assert; } dev->sram_pool = of_gen_pool_get(pdev->dev.of_node, "sram", 0); if (!dev->sram_pool) dev_warn(&pdev->dev, "sram node not found\n"); + dev->sram_size = match_data->sram_size; + dev->product_code = wave5_vdi_read_register(dev, VPU_PRODUCT_CODE_REGISTER); ret = wave5_vdi_init(&pdev->dev); if (ret < 0) { @@ -185,6 +266,28 @@ static int wave5_vpu_probe(struct platform_device *pdev) } dev->product = wave5_vpu_get_product_id(dev); + dev->irq = platform_get_irq(pdev, 0); + if (dev->irq < 0) { + dev_err(&pdev->dev, "failed to get irq resource, falling back to polling\n"); + hrtimer_setup(&dev->hrtimer, &wave5_vpu_timer_callback, CLOCK_MONOTONIC, + HRTIMER_MODE_REL_PINNED); + dev->worker = kthread_run_worker(0, "vpu_irq_thread"); + if (IS_ERR(dev->worker)) { + dev_err(&pdev->dev, "failed to create vpu irq worker\n"); + ret = PTR_ERR(dev->worker); + goto err_vdi_release; + } + dev->vpu_poll_interval = vpu_poll_interval; + kthread_init_work(&dev->work, wave5_vpu_irq_work_fn); + } else { + ret = devm_request_threaded_irq(&pdev->dev, dev->irq, NULL, + wave5_vpu_irq_thread, IRQF_ONESHOT, "vpu_irq", dev); + if (ret) { + dev_err(&pdev->dev, "Register interrupt handler, fail: %d\n", ret); + goto err_enc_unreg; + } + } + INIT_LIST_HEAD(&dev->instances); ret = v4l2_device_register(&pdev->dev, &dev->v4l2_dev); if (ret) { @@ -207,20 +310,6 @@ static int wave5_vpu_probe(struct platform_device *pdev) } } - dev->irq = platform_get_irq(pdev, 0); - if (dev->irq < 0) { - dev_err(&pdev->dev, "failed to get irq resource\n"); - ret = -ENXIO; - goto err_enc_unreg; - } - - ret = devm_request_threaded_irq(&pdev->dev, dev->irq, NULL, - wave5_vpu_irq_thread, IRQF_ONESHOT, "vpu_irq", dev); - if (ret) { - dev_err(&pdev->dev, "Register interrupt handler, fail: %d\n", ret); - goto err_enc_unreg; - } - ret = wave5_vpu_load_firmware(&pdev->dev, match_data->fw_name, &fw_revision); if (ret) { dev_err(&pdev->dev, "wave5_vpu_load_firmware, fail: %d\n", ret); @@ -232,6 +321,12 @@ static int wave5_vpu_probe(struct platform_device *pdev) (match_data->flags & WAVE5_IS_DEC) ? "'DECODE'" : ""); dev_info(&pdev->dev, "Product Code: 0x%x\n", dev->product_code); dev_info(&pdev->dev, "Firmware Revision: %u\n", fw_revision); + + pm_runtime_set_autosuspend_delay(&pdev->dev, 100); + pm_runtime_use_autosuspend(&pdev->dev); + pm_runtime_enable(&pdev->dev); + wave5_vpu_sleep_wake(&pdev->dev, true, NULL, 0); + return 0; err_enc_unreg: @@ -246,6 +341,8 @@ err_vdi_release: wave5_vdi_release(&pdev->dev); err_clk_dis: clk_bulk_disable_unprepare(dev->num_clks, dev->clks); +err_reset_assert: + reset_control_assert(dev->resets); return ret; } @@ -254,8 +351,17 @@ static void wave5_vpu_remove(struct platform_device *pdev) { struct vpu_device *dev = dev_get_drvdata(&pdev->dev); + if (dev->irq < 0) { + kthread_destroy_worker(dev->worker); + hrtimer_cancel(&dev->hrtimer); + } + + pm_runtime_put_sync(&pdev->dev); + pm_runtime_disable(&pdev->dev); + mutex_destroy(&dev->dev_lock); mutex_destroy(&dev->hw_lock); + reset_control_assert(dev->resets); clk_bulk_disable_unprepare(dev->num_clks, dev->clks); wave5_vpu_enc_unregister_device(dev); wave5_vpu_dec_unregister_device(dev); @@ -267,6 +373,7 @@ static void wave5_vpu_remove(struct platform_device *pdev) static const struct wave5_match_data ti_wave521c_data = { .flags = WAVE5_IS_ENC | WAVE5_IS_DEC, .fw_name = "cnm/wave521c_k3_codec_fw.bin", + .sram_size = (64 * 1024), }; static const struct of_device_id wave5_dt_ids[] = { @@ -279,9 +386,10 @@ static struct platform_driver wave5_vpu_driver = { .driver = { .name = VPU_PLATFORM_DEVICE_NAME, .of_match_table = of_match_ptr(wave5_dt_ids), + .pm = &wave5_pm_ops, }, .probe = wave5_vpu_probe, - .remove_new = wave5_vpu_remove, + .remove = wave5_vpu_remove, }; module_platform_driver(wave5_vpu_driver); diff --git a/drivers/media/platform/chips-media/wave5/wave5-vpu.h b/drivers/media/platform/chips-media/wave5/wave5-vpu.h index 32b7fd3730b5..3847332551fc 100644 --- a/drivers/media/platform/chips-media/wave5/wave5-vpu.h +++ b/drivers/media/platform/chips-media/wave5/wave5-vpu.h @@ -38,10 +38,7 @@ enum vpu_fmt_type { struct vpu_format { unsigned int v4l2_pix_fmt; - unsigned int max_width; - unsigned int min_width; - unsigned int max_height; - unsigned int min_height; + const struct v4l2_frmsize_stepwise *v4l2_frmsize; }; static inline struct vpu_instance *wave5_to_vpu_inst(struct v4l2_fh *vfh) diff --git a/drivers/media/platform/chips-media/wave5/wave5-vpuapi.c b/drivers/media/platform/chips-media/wave5/wave5-vpuapi.c index 1a3efb638dde..e5e879a13e8b 100644 --- a/drivers/media/platform/chips-media/wave5/wave5-vpuapi.c +++ b/drivers/media/platform/chips-media/wave5/wave5-vpuapi.c @@ -6,6 +6,8 @@ */ #include <linux/bug.h> +#include <linux/pm_runtime.h> +#include <linux/delay.h> #include "wave5-vpuapi.h" #include "wave5-regdefine.h" #include "wave5.h" @@ -73,6 +75,16 @@ int wave5_vpu_flush_instance(struct vpu_instance *inst) inst->type == VPU_INST_TYPE_DEC ? "DECODER" : "ENCODER", inst->id); mutex_unlock(&inst->dev->hw_lock); return -ETIMEDOUT; + } else if (ret == -EBUSY) { + struct dec_output_info dec_info; + + mutex_unlock(&inst->dev->hw_lock); + wave5_vpu_dec_get_output_info(inst, &dec_info); + ret = mutex_lock_interruptible(&inst->dev->hw_lock); + if (ret) + return ret; + if (dec_info.index_frame_display > 0) + wave5_vpu_dec_set_disp_flag(inst, dec_info.index_frame_display); } } while (ret != 0); mutex_unlock(&inst->dev->hw_lock); @@ -195,14 +207,20 @@ int wave5_vpu_dec_close(struct vpu_instance *inst, u32 *fail_res) int retry = 0; struct vpu_device *vpu_dev = inst->dev; int i; + int inst_count = 0; + struct vpu_instance *inst_elm; *fail_res = 0; if (!inst->codec_info) return -EINVAL; + pm_runtime_resume_and_get(inst->dev->dev); + ret = mutex_lock_interruptible(&vpu_dev->hw_lock); - if (ret) + if (ret) { + pm_runtime_put_sync(inst->dev->dev); return ret; + } do { ret = wave5_vpu_dec_finish_seq(inst, fail_res); @@ -232,9 +250,14 @@ int wave5_vpu_dec_close(struct vpu_instance *inst, u32 *fail_res) wave5_vdi_free_dma_memory(vpu_dev, &p_dec_info->vb_task); + list_for_each_entry(inst_elm, &vpu_dev->instances, list) + inst_count++; + if (inst_count == 1) + pm_runtime_dont_use_autosuspend(vpu_dev->dev); + unlock_and_return: mutex_unlock(&vpu_dev->hw_lock); - + pm_runtime_put_sync(inst->dev->dev); return ret; } @@ -697,25 +720,33 @@ int wave5_vpu_enc_close(struct vpu_instance *inst, u32 *fail_res) int ret; int retry = 0; struct vpu_device *vpu_dev = inst->dev; + int inst_count = 0; + struct vpu_instance *inst_elm; *fail_res = 0; if (!inst->codec_info) return -EINVAL; + pm_runtime_resume_and_get(inst->dev->dev); + ret = mutex_lock_interruptible(&vpu_dev->hw_lock); - if (ret) + if (ret) { + pm_runtime_resume_and_get(inst->dev->dev); return ret; + } do { ret = wave5_vpu_enc_finish_seq(inst, fail_res); if (ret < 0 && *fail_res != WAVE5_SYSERR_VPU_STILL_RUNNING) { dev_warn(inst->dev->dev, "enc_finish_seq timed out\n"); + pm_runtime_resume_and_get(inst->dev->dev); mutex_unlock(&vpu_dev->hw_lock); return ret; } if (*fail_res == WAVE5_SYSERR_VPU_STILL_RUNNING && retry++ >= MAX_FIRMWARE_CALL_RETRY) { + pm_runtime_resume_and_get(inst->dev->dev); mutex_unlock(&vpu_dev->hw_lock); return -ETIMEDOUT; } @@ -734,7 +765,13 @@ int wave5_vpu_enc_close(struct vpu_instance *inst, u32 *fail_res) wave5_vdi_free_dma_memory(vpu_dev, &p_enc_info->vb_task); + list_for_each_entry(inst_elm, &vpu_dev->instances, list) + inst_count++; + if (inst_count == 1) + pm_runtime_dont_use_autosuspend(vpu_dev->dev); + mutex_unlock(&vpu_dev->hw_lock); + pm_runtime_put_sync(inst->dev->dev); return 0; } diff --git a/drivers/media/platform/chips-media/wave5/wave5-vpuapi.h b/drivers/media/platform/chips-media/wave5/wave5-vpuapi.h index 352f6e904e50..45615c15beca 100644 --- a/drivers/media/platform/chips-media/wave5/wave5-vpuapi.h +++ b/drivers/media/platform/chips-media/wave5/wave5-vpuapi.h @@ -18,6 +18,7 @@ #include "wave5-vdi.h" enum product_id { + PRODUCT_ID_515, PRODUCT_ID_521, PRODUCT_ID_511, PRODUCT_ID_517, @@ -327,6 +328,7 @@ struct vpu_attr { u32 support_backbone: 1; u32 support_avc10bit_enc: 1; u32 support_hevc10bit_enc: 1; + u32 support_hevc10bit_dec: 1; u32 support_vcore_backbone: 1; u32 support_vcpu_backbone: 1; }; @@ -566,6 +568,7 @@ struct enc_wave_param { u32 lambda_scaling_enable: 1; /* enable lambda scaling using custom GOP */ u32 transform8x8_enable: 1; /* enable 8x8 intra prediction and 8x8 transform */ u32 mb_level_rc_enable: 1; /* enable MB-level rate control */ + u32 forced_idr_header_enable: 1; /* enable header encoding before IDR frame */ }; struct enc_open_param { @@ -756,7 +759,12 @@ struct vpu_device { u32 product_code; struct ida inst_ida; struct clk_bulk_data *clks; + struct hrtimer hrtimer; + struct kthread_work work; + struct kthread_worker *worker; + int vpu_poll_interval; int num_clks; + struct reset_control *resets; }; struct vpu_instance; diff --git a/drivers/media/platform/chips-media/wave5/wave5-vpuconfig.h b/drivers/media/platform/chips-media/wave5/wave5-vpuconfig.h index d9751eedb0f9..1ea9f5f31499 100644 --- a/drivers/media/platform/chips-media/wave5/wave5-vpuconfig.h +++ b/drivers/media/platform/chips-media/wave5/wave5-vpuconfig.h @@ -8,6 +8,7 @@ #ifndef _VPU_CONFIG_H_ #define _VPU_CONFIG_H_ +#define WAVE515_CODE 0x5150 #define WAVE517_CODE 0x5170 #define WAVE537_CODE 0x5370 #define WAVE511_CODE 0x5110 @@ -21,19 +22,39 @@ ((c) == WAVE517_CODE || (c) == WAVE537_CODE || \ (c) == WAVE511_CODE || (c) == WAVE521_CODE || \ (c) == WAVE521E1_CODE || (c) == WAVE521C_CODE || \ - (c) == WAVE521C_DUAL_CODE); \ + (c) == WAVE521C_DUAL_CODE) || (c) == WAVE515_CODE; \ }) #define WAVE517_WORKBUF_SIZE (2 * 1024 * 1024) #define WAVE521ENC_WORKBUF_SIZE (128 * 1024) //HEVC 128K, AVC 40K #define WAVE521DEC_WORKBUF_SIZE (1784 * 1024) +#define WAVE515DEC_WORKBUF_SIZE (2 * 1024 * 1024) #define MAX_NUM_INSTANCE 32 -#define W5_MIN_ENC_PIC_WIDTH 256 -#define W5_MIN_ENC_PIC_HEIGHT 128 -#define W5_MAX_ENC_PIC_WIDTH 8192 -#define W5_MAX_ENC_PIC_HEIGHT 8192 +#define W5_DEF_DEC_PIC_WIDTH 720U +#define W5_DEF_DEC_PIC_HEIGHT 480U +#define W5_MIN_DEC_PIC_8_WIDTH 8U +#define W5_MIN_DEC_PIC_8_HEIGHT 8U +#define W5_MIN_DEC_PIC_32_WIDTH 32U +#define W5_MIN_DEC_PIC_32_HEIGHT 32U +#define W5_MAX_DEC_PIC_WIDTH 8192U +#define W5_MAX_DEC_PIC_HEIGHT 4320U +#define W5_DEC_CODEC_STEP_WIDTH 1U +#define W5_DEC_CODEC_STEP_HEIGHT 1U +#define W5_DEC_RAW_STEP_WIDTH 32U +#define W5_DEC_RAW_STEP_HEIGHT 16U + +#define W5_DEF_ENC_PIC_WIDTH 416U +#define W5_DEF_ENC_PIC_HEIGHT 240U +#define W5_MIN_ENC_PIC_WIDTH 256U +#define W5_MIN_ENC_PIC_HEIGHT 128U +#define W5_MAX_ENC_PIC_WIDTH 8192U +#define W5_MAX_ENC_PIC_HEIGHT 8192U +#define W5_ENC_CODEC_STEP_WIDTH 8U +#define W5_ENC_CODEC_STEP_HEIGHT 8U +#define W5_ENC_RAW_STEP_WIDTH 32U +#define W5_ENC_RAW_STEP_HEIGHT 16U // application specific configuration #define VPU_ENC_TIMEOUT 60000 @@ -49,17 +70,21 @@ /************************************************************************/ #define VLC_BUF_NUM (2) -#define COMMAND_QUEUE_DEPTH (2) +#define WAVE521_COMMAND_QUEUE_DEPTH (2) +#define WAVE515_COMMAND_QUEUE_DEPTH (4) #define W5_REMAP_INDEX0 0 #define W5_REMAP_INDEX1 1 #define W5_REMAP_MAX_SIZE (1024 * 1024) -#define WAVE5_MAX_CODE_BUF_SIZE (2 * 1024 * 1024) -#define WAVE5_TEMPBUF_OFFSET WAVE5_MAX_CODE_BUF_SIZE +#define WAVE521_MAX_CODE_BUF_SIZE (2 * 1024 * 1024) +#define WAVE515_MAX_CODE_BUF_SIZE (1024 * 1024) #define WAVE5_TEMPBUF_SIZE (1024 * 1024) -#define SIZE_COMMON (WAVE5_MAX_CODE_BUF_SIZE + WAVE5_TEMPBUF_SIZE) +#define WAVE521_SIZE_COMMON (WAVE521_MAX_CODE_BUF_SIZE + WAVE5_TEMPBUF_SIZE) +#define WAVE515_ONE_TASKBUF_SIZE (8 * 1024 * 1024) +#define WAVE515_SIZE_COMMON (WAVE515_MAX_CODE_BUF_SIZE + WAVE5_TEMPBUF_SIZE + \ + WAVE515_COMMAND_QUEUE_DEPTH * WAVE515_ONE_TASKBUF_SIZE) //=====4. VPU REPORT MEMORY ======================// diff --git a/drivers/media/platform/chips-media/wave5/wave5.h b/drivers/media/platform/chips-media/wave5/wave5.h index 063028eccd3b..2caab356f3e1 100644 --- a/drivers/media/platform/chips-media/wave5/wave5.h +++ b/drivers/media/platform/chips-media/wave5/wave5.h @@ -22,6 +22,12 @@ */ #define BSOPTION_ENABLE_EXPLICIT_END BIT(0) #define BSOPTION_HIGHLIGHT_STREAM_END BIT(1) +/* + * When RD_PTR_VALID_FLAG is 0 Wave515 ignores RD_PTR value and starts to + * decode from the access unit end position of the last decoded picture in + * bitstream buffer. + */ +#define BSOPTION_RD_PTR_VALID_FLAG BIT(31) /* * Currently the driver only supports hardware with little endian but for source @@ -56,6 +62,9 @@ int wave5_vpu_get_version(struct vpu_device *vpu_dev, u32 *revision); int wave5_vpu_init(struct device *dev, u8 *fw, size_t size); +int wave5_vpu_sleep_wake(struct device *dev, bool i_sleep_wake, const uint16_t *code, + size_t size); + int wave5_vpu_reset(struct device *dev, enum sw_reset_mode reset_mode); int wave5_vpu_build_up_dec_param(struct vpu_instance *inst, struct dec_open_param *param); |