Re: [RFC PATCH v6] media: mediatek: vcodec: support stateless AV1 decoder
From: Andrzej Pietrasiewicz
Date: Thu Nov 17 2022 - 07:46:14 EST
Hi Xiaoyong Lu,
Sorry about chiming in only at v6. Please see inline below.
Andrzej
W dniu 17.11.2022 o 07:17, Xiaoyong Lu pisze:
Add mediatek av1 decoder linux driver which use the stateless API in
MT8195.
Signed-off-by: Xiaoyong Lu<xiaoyong.lu@xxxxxxxxxxxx>
---
Changes from v5:
- change av1 PROFILE and LEVEL cfg
- test by av1 fluster, result is 173/239
Changes from v4:
- convert vb2_find_timestamp to vb2_find_buffer
- test by av1 fluster, result is 173/239
Changes from v3:
- modify comment for struct vdec_av1_slice_slot
- add define SEG_LVL_ALT_Q
- change use_lr/use_chroma_lr parse from av1 spec
- use ARRAY_SIZE to replace size for loop_filter_level and loop_filter_mode_deltas
- change array size of loop_filter_mode_deltas from 4 to 2
- add define SECONDARY_FILTER_STRENGTH_NUM_BITS
- change some hex values from upper case to lower case
- change *dpb_sz equal to V4L2_AV1_TOTAL_REFS_PER_FRAME + 1
- test by av1 fluster, result is 173/239
Changes from v2:
- Match with av1 uapi v3 modify
- test by av1 fluster, result is 173/239
---
Reference series:
[1]: v3 of this series is presend by Daniel Almeida.
message-id: 20220825225312.564619-1-daniel.almeida@xxxxxxxxxxxxx
.../media/platform/mediatek/vcodec/Makefile | 1 +
.../vcodec/mtk_vcodec_dec_stateless.c | 47 +-
.../platform/mediatek/vcodec/mtk_vcodec_drv.h | 1 +
.../vcodec/vdec/vdec_av1_req_lat_if.c | 2234 +++++++++++++++++
.../platform/mediatek/vcodec/vdec_drv_if.c | 4 +
.../platform/mediatek/vcodec/vdec_drv_if.h | 1 +
.../platform/mediatek/vcodec/vdec_msg_queue.c | 27 +
.../platform/mediatek/vcodec/vdec_msg_queue.h | 4 +
8 files changed, 2318 insertions(+), 1 deletion(-)
create mode 100644 drivers/media/platform/mediatek/vcodec/vdec/vdec_av1_req_lat_if.c
diff --git a/drivers/media/platform/mediatek/vcodec/Makefile b/drivers/media/platform/mediatek/vcodec/Makefile
index 93e7a343b5b0e..7537259130072 100644
--- a/drivers/media/platform/mediatek/vcodec/Makefile
+++ b/drivers/media/platform/mediatek/vcodec/Makefile
@@ -10,6 +10,7 @@ mtk-vcodec-dec-y := vdec/vdec_h264_if.o \
vdec/vdec_vp8_req_if.o \
vdec/vdec_vp9_if.o \
vdec/vdec_vp9_req_lat_if.o \
+ vdec/vdec_av1_req_lat_if.o \
vdec/vdec_h264_req_if.o \
vdec/vdec_h264_req_common.o \
vdec/vdec_h264_req_multi_if.o \
diff --git a/drivers/media/platform/mediatek/vcodec/mtk_vcodec_dec_stateless.c b/drivers/media/platform/mediatek/vcodec/mtk_vcodec_dec_stateless.c
index c45bd2599bb2d..ceb6fabc67749 100644
--- a/drivers/media/platform/mediatek/vcodec/mtk_vcodec_dec_stateless.c
+++ b/drivers/media/platform/mediatek/vcodec/mtk_vcodec_dec_stateless.c
@@ -107,11 +107,51 @@ static const struct mtk_stateless_control mtk_stateless_controls[] = {
},
.codec_type = V4L2_PIX_FMT_VP9_FRAME,
},
+ {
+ .cfg = {
+ .id = V4L2_CID_STATELESS_AV1_SEQUENCE,
+
+ },
+ .codec_type = V4L2_PIX_FMT_AV1_FRAME,
+ },
+ {
+ .cfg = {
+ .id = V4L2_CID_STATELESS_AV1_FRAME,
+
+ },
+ .codec_type = V4L2_PIX_FMT_AV1_FRAME,
+ },
+ {
+ .cfg = {
+ .id = V4L2_CID_STATELESS_AV1_TILE_GROUP_ENTRY,
+ .dims = { V4L2_AV1_MAX_TILE_COUNT },
+
+ },
+ .codec_type = V4L2_PIX_FMT_AV1_FRAME,
+ },
+ {
+ .cfg = {
+ .id = V4L2_CID_STATELESS_AV1_PROFILE,
+ .min = V4L2_STATELESS_AV1_PROFILE_MAIN,
+ .def = V4L2_STATELESS_AV1_PROFILE_MAIN,
+ .max = V4L2_STATELESS_AV1_PROFILE_MAIN,
+ },
+ .codec_type = V4L2_PIX_FMT_AV1_FRAME,
+ },
+ {
+ .cfg = {
+ .id = V4L2_CID_STATELESS_AV1_LEVEL,
+ .min = V4L2_STATELESS_AV1_LEVEL_2_0,
+ .def = V4L2_STATELESS_AV1_LEVEL_4_0,
+ .max = V4L2_STATELESS_AV1_LEVEL_5_1,
+ },
+ .codec_type = V4L2_PIX_FMT_AV1_FRAME,
+ },
};
#define NUM_CTRLS ARRAY_SIZE(mtk_stateless_controls)
-static struct mtk_video_fmt mtk_video_formats[5];
+static struct mtk_video_fmt mtk_video_formats[6];
static struct mtk_video_fmt default_out_format;
static struct mtk_video_fmt default_cap_format;
@@ -351,6 +391,7 @@ static void mtk_vcodec_add_formats(unsigned int fourcc,
case V4L2_PIX_FMT_H264_SLICE:
case V4L2_PIX_FMT_VP8_FRAME:
case V4L2_PIX_FMT_VP9_FRAME:
+ case V4L2_PIX_FMT_AV1_FRAME:
mtk_video_formats[count_formats].fourcc = fourcc;
mtk_video_formats[count_formats].type = MTK_FMT_DEC;
mtk_video_formats[count_formats].num_planes = 1;
@@ -407,6 +448,10 @@ static void mtk_vcodec_get_supported_formats(struct mtk_vcodec_ctx *ctx)
mtk_vcodec_add_formats(V4L2_PIX_FMT_VP9_FRAME, ctx);
out_format_count++;
}
+ if (ctx->dev->dec_capability & MTK_VDEC_FORMAT_AV1_FRAME) {
+ mtk_vcodec_add_formats(V4L2_PIX_FMT_AV1_FRAME, ctx);
+ out_format_count++;
+ }
if (cap_format_count)
default_cap_format = mtk_video_formats[cap_format_count - 1];
diff --git a/drivers/media/platform/mediatek/vcodec/mtk_vcodec_drv.h b/drivers/media/platform/mediatek/vcodec/mtk_vcodec_drv.h
index 6a47a11ff654a..a6db972b1ff72 100644
--- a/drivers/media/platform/mediatek/vcodec/mtk_vcodec_drv.h
+++ b/drivers/media/platform/mediatek/vcodec/mtk_vcodec_drv.h
@@ -344,6 +344,7 @@ enum mtk_vdec_format_types {
MTK_VDEC_FORMAT_H264_SLICE = 0x100,
MTK_VDEC_FORMAT_VP8_FRAME = 0x200,
MTK_VDEC_FORMAT_VP9_FRAME = 0x400,
+ MTK_VDEC_FORMAT_AV1_FRAME = 0x800,
MTK_VCODEC_INNER_RACING = 0x20000,
};
diff --git a/drivers/media/platform/mediatek/vcodec/vdec/vdec_av1_req_lat_if.c b/drivers/media/platform/mediatek/vcodec/vdec/vdec_av1_req_lat_if.c
new file mode 100644
index 0000000000000..2ac77175dad7c
--- /dev/null
+++ b/drivers/media/platform/mediatek/vcodec/vdec/vdec_av1_req_lat_if.c
@@ -0,0 +1,2234 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2022 MediaTek Inc.
+ * Author: Xiaoyong Lu <xiaoyong.lu@xxxxxxxxxxxx>
+ */
+
+#include <linux/module.h>
+#include <linux/slab.h>
+#include <media/videobuf2-dma-contig.h>
+
+#include "../mtk_vcodec_util.h"
+#include "../mtk_vcodec_dec.h"
+#include "../mtk_vcodec_intr.h"
+#include "../vdec_drv_base.h"
+#include "../vdec_drv_if.h"
+#include "../vdec_vpu_if.h"
+
+#define AV1_MAX_FRAME_BUF_COUNT (V4L2_AV1_TOTAL_REFS_PER_FRAME + 1)
+#define AV1_TILE_BUF_SIZE 64
+#define AV1_SCALE_SUBPEL_BITS 10
+#define AV1_REF_SCALE_SHIFT 14
+#define AV1_REF_NO_SCALE BIT(AV1_REF_SCALE_SHIFT)
+#define AV1_REF_INVALID_SCALE -1
+
+#define AV1_INVALID_IDX -1
+
+#define AV1_DIV_ROUND_UP_POW2(value, n) \
+({ \
+ typeof(n) _n = n; \
+ typeof(value) _value = value; \
+ (_value + (BIT(_n) >> 1)) >> _n; \
+})
+
+#define AV1_DIV_ROUND_UP_POW2_SIGNED(value, n) \
+({ \
+ typeof(n) _n_ = n; \
+ typeof(value) _value_ = value; \
+ (((_value_) < 0) ? -AV1_DIV_ROUND_UP_POW2(-(_value_), (_n_)) \
+ : AV1_DIV_ROUND_UP_POW2((_value_), (_n_))); \
+})
+
+#define BIT_FLAG(x, bit) (!!((x)->flags & (bit)))
+#define SEGMENTATION_FLAG(x, name) (!!((x)->flags & V4L2_AV1_SEGMENTATION_FLAG_##name))
+#define QUANT_FLAG(x, name) (!!((x)->flags & V4L2_AV1_QUANTIZATION_FLAG_##name))
+#define SEQUENCE_FLAG(x, name) (!!((x)->flags & V4L2_AV1_SEQUENCE_FLAG_##name))
+#define FH_FLAG(x, name) (!!((x)->flags & V4L2_AV1_FRAME_FLAG_##name))
+
+#define MINQ 0
+#define MAXQ 255
+
+#define DIV_LUT_PREC_BITS 14
+#define DIV_LUT_BITS 8
+#define DIV_LUT_NUM BIT(DIV_LUT_BITS)
+#define WARP_PARAM_REDUCE_BITS 6
+#define WARPEDMODEL_PREC_BITS 16
+
+#define SEG_LVL_ALT_Q 0
+#define SECONDARY_FILTER_STRENGTH_NUM_BITS 2
+
+static const short div_lut[DIV_LUT_NUM + 1] = {
+ 16384, 16320, 16257, 16194, 16132, 16070, 16009, 15948, 15888, 15828, 15768,
+ 15709, 15650, 15592, 15534, 15477, 15420, 15364, 15308, 15252, 15197, 15142,
+ 15087, 15033, 14980, 14926, 14873, 14821, 14769, 14717, 14665, 14614, 14564,
+ 14513, 14463, 14413, 14364, 14315, 14266, 14218, 14170, 14122, 14075, 14028,
+ 13981, 13935, 13888, 13843, 13797, 13752, 13707, 13662, 13618, 13574, 13530,
+ 13487, 13443, 13400, 13358, 13315, 13273, 13231, 13190, 13148, 13107, 13066,
+ 13026, 12985, 12945, 12906, 12866, 12827, 12788, 12749, 12710, 12672, 12633,
+ 12596, 12558, 12520, 12483, 12446, 12409, 12373, 12336, 12300, 12264, 12228,
+ 12193, 12157, 12122, 12087, 12053, 12018, 11984, 11950, 11916, 11882, 11848,
+ 11815, 11782, 11749, 11716, 11683, 11651, 11619, 11586, 11555, 11523, 11491,
+ 11460, 11429, 11398, 11367, 11336, 11305, 11275, 11245, 11215, 11185, 11155,
+ 11125, 11096, 11067, 11038, 11009, 10980, 10951, 10923, 10894, 10866, 10838,
+ 10810, 10782, 10755, 10727, 10700, 10673, 10645, 10618, 10592, 10565, 10538,
+ 10512, 10486, 10460, 10434, 10408, 10382, 10356, 10331, 10305, 10280, 10255,
+ 10230, 10205, 10180, 10156, 10131, 10107, 10082, 10058, 10034, 10010, 9986,
+ 9963, 9939, 9916, 9892, 9869, 9846, 9823, 9800, 9777, 9754, 9732,
+ 9709, 9687, 9664, 9642, 9620, 9598, 9576, 9554, 9533, 9511, 9489,
+ 9468, 9447, 9425, 9404, 9383, 9362, 9341, 9321, 9300, 9279, 9259,
+ 9239, 9218, 9198, 9178, 9158, 9138, 9118, 9098, 9079, 9059, 9039,
+ 9020, 9001, 8981, 8962, 8943, 8924, 8905, 8886, 8867, 8849, 8830,
+ 8812, 8793, 8775, 8756, 8738, 8720, 8702, 8684, 8666, 8648, 8630,
+ 8613, 8595, 8577, 8560, 8542, 8525, 8508, 8490, 8473, 8456, 8439,
+ 8422, 8405, 8389, 8372, 8355, 8339, 8322, 8306, 8289, 8273, 8257,
+ 8240, 8224, 8208, 8192,
+};
+
+/**
+ * struct vdec_av1_slice_init_vsi - VSI used to initialize instance
+ * @architecture: architecture type
+ * @reserved: reserved
+ * @core_vsi: for core vsi
+ * @cdf_table_addr: cdf table addr
+ * @cdf_table_size: cdf table size
+ * @iq_table_addr: iq table addr
+ * @iq_table_size: iq table size
+ * @vsi_size: share vsi structure size
+ */
+struct vdec_av1_slice_init_vsi {
+ u32 architecture;
+ u32 reserved;
+ u64 core_vsi;
+ u64 cdf_table_addr;
+ u32 cdf_table_size;
+ u64 iq_table_addr;
+ u32 iq_table_size;
+ u32 vsi_size;
+};
+
+/**
+ * struct vdec_av1_slice_mem - memory address and size
+ * @buf: dma_addr padding
+ * @dma_addr: buffer address
+ * @size: buffer size
+ * @dma_addr_end: buffer end address
+ * @padding: for padding
+ */
+struct vdec_av1_slice_mem {
+ union {
+ u64 buf;
+ dma_addr_t dma_addr;
+ };
+ union {
+ size_t size;
+ dma_addr_t dma_addr_end;
+ u64 padding;
+ };
+};
+
+/**
+ * struct vdec_av1_slice_state - decoding state
+ * @err : err type for decode
+ * @full : transcoded buffer is full or not
+ * @timeout : decode timeout or not
+ * @perf : performance enable
+ * @crc : hw checksum
+ * @out_size : hw output size
+ */
+struct vdec_av1_slice_state {
+ int err;
+ u32 full;
+ u32 timeout;
+ u32 perf;
+ u32 crc[16];
+ u32 out_size;
+};
+
+/*
+ * enum vdec_av1_slice_resolution_level - resolution level
+ */
+enum vdec_av1_slice_resolution_level {
+ AV1_RES_NONE,
+ AV1_RES_FHD,
+ AV1_RES_4K,
+ AV1_RES_8K,
+};
+
+/*
+ * enum vdec_av1_slice_frame_type - av1 frame type
+ */
+enum vdec_av1_slice_frame_type {
+ AV1_KEY_FRAME = 0,
+ AV1_INTER_FRAME,
+ AV1_INTRA_ONLY_FRAME,
+ AV1_SWITCH_FRAME,
+ AV1_FRAME_TYPES,
+};
+
+/*
+ * enum vdec_av1_slice_reference_mode - reference mode type
+ */
+enum vdec_av1_slice_reference_mode {
+ AV1_SINGLE_REFERENCE = 0,
+ AV1_COMPOUND_REFERENCE,
+ AV1_REFERENCE_MODE_SELECT,
+ AV1_REFERENCE_MODES,
+};
+
+/**
+ * struct vdec_av1_slice_tile_group - info for each tile
+ * @num_tiles: tile number
+ * @tile_size: input size for each tile
+ * @tile_start_offset: tile offset to input buffer
+ */
+struct vdec_av1_slice_tile_group {
+ u32 num_tiles;
+ u32 tile_size[V4L2_AV1_MAX_TILE_COUNT];
+ u32 tile_start_offset[V4L2_AV1_MAX_TILE_COUNT];
+};
+
+/**
+ * struct vdec_av1_slice_scale_factors - scale info for each ref frame
+ * @is_scaled: frame is scaled or not
+ * @x_scale: frame width scale coefficient
+ * @y_scale: frame height scale coefficient
+ * @x_step: width step for x_scale
+ * @y_step: height step for y_scale
+ */
+struct vdec_av1_slice_scale_factors {
+ u8 is_scaled;
+ int x_scale;
+ int y_scale;
+ int x_step;
+ int y_step;
+};
+
+/**
+ * struct vdec_av1_slice_frame_refs - ref frame info
+ * @ref_fb_idx: ref slot index
+ * @ref_map_idx: ref frame index
+ * @scale_factors: scale factors for each ref frame
+ */
+struct vdec_av1_slice_frame_refs {
+ int ref_fb_idx;
+ int ref_map_idx;
+ struct vdec_av1_slice_scale_factors scale_factors;
+};
+
+/**
+ * struct vdec_av1_slice_gm - AV1 Global Motion parameters
+ * @wmtype: The type of global motion transform used
+ * @wmmat: gm_params
+ * @alpha: alpha info
+ * @beta: beta info
+ * @gamma: gamma info
+ * @delta: delta info
+ * @invalid: is invalid or not
+ */
+struct vdec_av1_slice_gm {
+ int wmtype;
+ int wmmat[8];
+ short alpha;
+ short beta;
+ short gamma;
+ short delta;
+ char invalid;
+};
+
+/**
+ * struct vdec_av1_slice_sm - AV1 Skip Mode parameters
+ * @skip_mode_allowed: Skip Mode is allowed or not
+ * @skip_mode_present: specified that the skip_mode will be present or not
+ * @skip_mode_frame: specifies the frames to use for compound prediction
+ */
+struct vdec_av1_slice_sm {
+ u8 skip_mode_allowed;
+ u8 skip_mode_present;
+ int skip_mode_frame[2];
+};
+
+/**
+ * struct vdec_av1_slice_seg - AV1 Segmentation params
+ * @segmentation_enabled: this frame makes use of the segmentation tool or not
+ * @segmentation_update_map: segmentation map are updated during the decoding frame
+ * @segmentation_temporal_update:segmentation map are coded relative the existing segmentaion map
+ * @segmentation_update_data: new parameters are about to be specified for each segment
+ * @feature_data: specifies the feature data for a segment feature
+ * @feature_enabled_mask: the corresponding feature value is coded or not.
+ * @segid_preskip: segment id will be read before the skip syntax element.
+ * @last_active_segid: the highest numbered segment id that has some enabled feature
+ */
+struct vdec_av1_slice_seg {
+ u8 segmentation_enabled;
+ u8 segmentation_update_map;
+ u8 segmentation_temporal_update;
+ u8 segmentation_update_data;
+ int feature_data[V4L2_AV1_MAX_SEGMENTS][V4L2_AV1_SEG_LVL_MAX];
+ u16 feature_enabled_mask[V4L2_AV1_MAX_SEGMENTS];
+ int segid_preskip;
+ int last_active_segid;
+};
+
+/**
+ * struct vdec_av1_slice_delta_q_lf - AV1 Loop Filter delta parameters
+ * @delta_q_present: specified whether quantizer index delta values are present
+ * @delta_q_res: specifies the left shift which should be applied to decoded quantizer index
+ * @delta_lf_present: specifies whether loop filter delta values are present
+ * @delta_lf_res: specifies the left shift which should be applied to decoded
+ * loop filter delta values
+ * @delta_lf_multi: specifies that separate loop filter deltas are sent for horizontal
+ * luma edges,vertical luma edges,the u edges, and the v edges.
+ */
+struct vdec_av1_slice_delta_q_lf {
+ u8 delta_q_present;
+ u8 delta_q_res;
+ u8 delta_lf_present;
+ u8 delta_lf_res;
+ u8 delta_lf_multi;
+};
+
+/**
+ * struct vdec_av1_slice_quantization - AV1 Quantization params
+ * @base_q_idx: indicates the base frame qindex. This is used for Y AC
+ * coefficients and as the base value for the other quantizers.
+ * @qindex: qindex
+ * @delta_qydc: indicates the Y DC quantizer relative to base_q_idx
+ * @delta_qudc: indicates the U DC quantizer relative to base_q_idx.
+ * @delta_quac: indicates the U AC quantizer relative to base_q_idx
+ * @delta_qvdc: indicates the V DC quantizer relative to base_q_idx
+ * @delta_qvac: indicates the V AC quantizer relative to base_q_idx
+ * @using_qmatrix: specifies that the quantizer matrix will be used to
+ * compute quantizers
+ * @qm_y: specifies the level in the quantizer matrix that should
+ * be used for luma plane decoding
+ * @qm_u: specifies the level in the quantizer matrix that should
+ * be used for chroma U plane decoding.
+ * @qm_v: specifies the level in the quantizer matrix that should be
+ * used for chroma V plane decoding
+ */
+struct vdec_av1_slice_quantization {
+ int base_q_idx;
+ int qindex[V4L2_AV1_MAX_SEGMENTS];
+ int delta_qydc;
+ int delta_qudc;
+ int delta_quac;
+ int delta_qvdc;
+ int delta_qvac;
+ u8 using_qmatrix;
+ u8 qm_y;
+ u8 qm_u;
+ u8 qm_v;
+};
+
+/**
+ * struct vdec_av1_slice_lr - AV1 Loop Restauration parameters
+ * @use_lr: whether to use loop restoration
+ * @use_chroma_lr: whether to use chroma loop restoration
+ * @frame_restoration_type: specifies the type of restoration used for each plane
+ * @loop_restoration_size: pecifies the size of loop restoration units in units
+ * of samples in the current plane
+ */
+struct vdec_av1_slice_lr {
+ u8 use_lr;
+ u8 use_chroma_lr;
+ u8 frame_restoration_type[V4L2_AV1_NUM_PLANES_MAX];
+ u32 loop_restoration_size[V4L2_AV1_NUM_PLANES_MAX];
+};
+
+/**
+ * struct vdec_av1_slice_loop_filter - AV1 Loop filter parameters
+ * @loop_filter_level: an array containing loop filter strength values.
+ * @loop_filter_ref_deltas: contains the adjustment needed for the filter
+ * level based on the chosen reference frame
+ * @loop_filter_mode_deltas: contains the adjustment needed for the filter
+ * level based on the chosen mode
+ * @loop_filter_sharpness: indicates the sharpness level. The loop_filter_level
+ * and loop_filter_sharpness together determine when
+ * a block edge is filtered, and by how much the
+ * filtering can change the sample values
+ * @loop_filter_delta_enabled: filetr level depends on the mode and reference
+ * frame used to predict a block
+ */
+struct vdec_av1_slice_loop_filter {
+ u8 loop_filter_level[4];
+ int loop_filter_ref_deltas[V4L2_AV1_TOTAL_REFS_PER_FRAME];
+ int loop_filter_mode_deltas[2];
+ u8 loop_filter_sharpness;
+ u8 loop_filter_delta_enabled;
+};
+
+/**
+ * struct vdec_av1_slice_cdef - AV1 CDEF parameters
+ * @cdef_damping: controls the amount of damping in the deringing filter
+ * @cdef_y_strength: specifies the strength of the primary filter and secondary filter
+ * @cdef_uv_strength: specifies the strength of the primary filter and secondary filter
+ * @cdef_bits: specifies the number of bits needed to specify which
+ * CDEF filter to apply
+ */
+struct vdec_av1_slice_cdef {
+ u8 cdef_damping;
+ u8 cdef_y_strength[8];
+ u8 cdef_uv_strength[8];
+ u8 cdef_bits;
+};
+
+/**
+ * struct vdec_av1_slice_mfmv - AV1 mfmv parameters
+ * @mfmv_valid_ref: mfmv_valid_ref
+ * @mfmv_dir: mfmv_dir
+ * @mfmv_ref_to_cur: mfmv_ref_to_cur
+ * @mfmv_ref_frame_idx: mfmv_ref_frame_idx
+ * @mfmv_count: mfmv_count
+ */
+struct vdec_av1_slice_mfmv {
+ u32 mfmv_valid_ref[3];
+ u32 mfmv_dir[3];
+ int mfmv_ref_to_cur[3];
+ int mfmv_ref_frame_idx[3];
+ int mfmv_count;
+};
+
+/**
+ * struct vdec_av1_slice_tile - AV1 Tile info
+ * @tile_cols: specifies the number of tiles across the frame
+ * @tile_rows: pecifies the number of tiles down the frame
+ * @mi_col_starts: an array specifying the start column
+ * @mi_row_starts: an array specifying the start row
+ * @context_update_tile_id: specifies which tile to use for the CDF update
+ * @uniform_tile_spacing_flag: tiles are uniformly spaced across the frame
+ * or the tile sizes are coded
+ */
+struct vdec_av1_slice_tile {
+ u8 tile_cols;
+ u8 tile_rows;
+ int mi_col_starts[V4L2_AV1_MAX_TILE_COLS + 1];
+ int mi_row_starts[V4L2_AV1_MAX_TILE_ROWS + 1];
+ u8 context_update_tile_id;
+ u8 uniform_tile_spacing_flag;
+};
+
+/**
+ * struct vdec_av1_slice_uncompressed_header - Represents an AV1 Frame Header OBU
+ * @use_ref_frame_mvs: use_ref_frame_mvs flag
+ * @order_hint: specifies OrderHintBits least significant bits of the expected
+ * @gm: global motion param
+ * @upscaled_width: the upscaled width
+ * @frame_width: frame's width
+ * @frame_height: frame's height
+ * @reduced_tx_set: frame is restricted to a reduced subset of the full
+ * set of transform types
+ * @tx_mode: specifies how the transform size is determined
+ * @uniform_tile_spacing_flag: tiles are uniformly spaced across the frame
+ * or the tile sizes are coded
+ * @interpolation_filter: specifies the filter selection used for performing inter prediction
+ * @allow_warped_motion: motion_mode may be present or not
+ * @is_motion_mode_switchable : euqlt to 0 specifies that only the SIMPLE motion mode will be used
+ * @reference_mode : frame reference mode selected
+ * @allow_high_precision_mv: specifies that motion vectors are specified to
+ * quarter pel precision or to eighth pel precision
+ * @allow_intra_bc: ubducates that intra block copy may be used in this frame
+ * @force_integer_mv: specifies motion vectors will always be integers or
+ * can contain fractional bits
+ * @allow_screen_content_tools: intra blocks may use palette encoding
+ * @error_resilient_mode: error resislent mode is enable/disable
+ * @frame_type: specifies the AV1 frame type
+ * @primary_ref_frame: specifies which reference frame contains the CDF values
+ * and other state that should be loaded at the start of the frame
+ * slots will be updated with the current frame after it is decoded
+ * @disable_frame_end_update_cdf:indicates the end of frame CDF update is disable or enable
+ * @disable_cdf_update: specified whether the CDF update in the symbol
+ * decoding process should be disables
+ * @skip_mode: av1 skip mode parameters
+ * @seg: av1 segmentaon parameters
+ * @delta_q_lf: av1 delta loop fileter
+ * @quant: av1 Quantization params
+ * @lr: av1 Loop Restauration parameters
+ * @superres_denom: the denominator for the upscaling ratio
+ * @loop_filter: av1 Loop filter parameters
+ * @cdef: av1 CDEF parameters
+ * @mfmv: av1 mfmv parameters
+ * @tile: av1 Tile info
+ * @frame_is_intra: intra frame
+ * @loss_less_array: loss less array
+ * @coded_loss_less: coded lsss less
+ * @mi_rows: size of mi unit in rows
+ * @mi_cols: size of mi unit in cols
+ */
+struct vdec_av1_slice_uncompressed_header {
+ u8 use_ref_frame_mvs;
+ int order_hint;
+ struct vdec_av1_slice_gm gm[V4L2_AV1_TOTAL_REFS_PER_FRAME];
+ u32 upscaled_width;
+ u32 frame_width;
+ u32 frame_height;
+ u8 reduced_tx_set;
+ u8 tx_mode;
+ u8 uniform_tile_spacing_flag;
+ u8 interpolation_filter;
+ u8 allow_warped_motion;
+ u8 is_motion_mode_switchable;
+ u8 reference_mode;
+ u8 allow_high_precision_mv;
+ u8 allow_intra_bc;
+ u8 force_integer_mv;
+ u8 allow_screen_content_tools;
+ u8 error_resilient_mode;
+ u8 frame_type;
+ u8 primary_ref_frame;
+ u8 disable_frame_end_update_cdf;
+ u32 disable_cdf_update;
+ struct vdec_av1_slice_sm skip_mode;
+ struct vdec_av1_slice_seg seg;
+ struct vdec_av1_slice_delta_q_lf delta_q_lf;
+ struct vdec_av1_slice_quantization quant;
+ struct vdec_av1_slice_lr lr;
+ u32 superres_denom;
+ struct vdec_av1_slice_loop_filter loop_filter;
+ struct vdec_av1_slice_cdef cdef;
+ struct vdec_av1_slice_mfmv mfmv;
+ struct vdec_av1_slice_tile tile;
+ u8 frame_is_intra;
+ u8 loss_less_array[V4L2_AV1_MAX_SEGMENTS];
+ u8 coded_loss_less;
+ u32 mi_rows;
+ u32 mi_cols;
+};
+
+/**
+ * struct vdec_av1_slice_seq_header - Represents an AV1 Sequence OBU
+ * @bitdepth: the bitdepth to use for the sequence
+ * @enable_superres: specifies whether the use_superres syntax element may be present
+ * @enable_filter_intra: specifies the use_filter_intra syntax element may be present
+ * @enable_intra_edge_filter: whether the intra edge filtering process should be enabled
+ * @enable_interintra_compound: specifies the mode info fo rinter blocks may
+ * contain the syntax element interintra
+ * @enable_masked_compound: specifies the mode info fo rinter blocks may
+ * contain the syntax element compound_type
+ * @enable_dual_filter: the inter prediction filter type may be specified independently
+ * @enable_jnt_comp: distance weights process may be used for inter prediction
+ * @mono_chrome: indicates the video does not contain U and V color planes
+ * @enable_order_hint: tools based on the values of order hints may be used
+ * @order_hint_bits: the number of bits used for the order_hint field at each frame
+ * @use_128x128_superblock: indicates superblocks contain 128*128 luma samples
+ * @subsampling_x: the chroma subsamling format
+ * @subsampling_y: the chroma subsamling format
+ * @max_frame_width: the maximum frame width for the frames represented by sequence
+ * @max_frame_height: the maximum frame height for the frames represented by sequence
+ */
+struct vdec_av1_slice_seq_header {
+ u8 bitdepth;
+ u8 enable_superres;
+ u8 enable_filter_intra;
+ u8 enable_intra_edge_filter;
+ u8 enable_interintra_compound;
+ u8 enable_masked_compound;
+ u8 enable_dual_filter;
+ u8 enable_jnt_comp;
+ u8 mono_chrome;
+ u8 enable_order_hint;
+ u8 order_hint_bits;
+ u8 use_128x128_superblock;
+ u8 subsampling_x;
+ u8 subsampling_y;
+ u32 max_frame_width;
+ u32 max_frame_height;
+};
+
+/**
+ * struct vdec_av1_slice_frame - Represents current Frame info
+ * @uh: uncompressed header info
+ * @seq: sequence header info
+ * @large_scale_tile: is large scale mode
+ * @cur_ts: current frame timestamp
+ * @prev_fb_idx: prev slot id
+ * @ref_frame_sign_bias: arrays for ref_frame sign bias
+ * @order_hints: arrays for ref_frame order hint
+ * @ref_frame_valid: arrays for valid ref_frame
+ * @ref_frame_map: map to slot frame info
+ * @frame_refs: ref_frame info
+ */
+struct vdec_av1_slice_frame {
+ struct vdec_av1_slice_uncompressed_header uh;
+ struct vdec_av1_slice_seq_header seq;
+ u8 large_scale_tile;
+ u64 cur_ts;
+ int prev_fb_idx;
+ u8 ref_frame_sign_bias[V4L2_AV1_TOTAL_REFS_PER_FRAME];
+ u32 order_hints[V4L2_AV1_REFS_PER_FRAME];
+ u32 ref_frame_valid[V4L2_AV1_REFS_PER_FRAME];
+ int ref_frame_map[V4L2_AV1_TOTAL_REFS_PER_FRAME];
+ struct vdec_av1_slice_frame_refs frame_refs[V4L2_AV1_REFS_PER_FRAME];
+};
+
+/**
+ * struct vdec_av1_slice_work_buffer - work buffer for lat
+ * @mv_addr: mv buffer memory info
+ * @cdf_addr: cdf buffer memory info
+ * @segid_addr: segid buffer memory info
+ */
+struct vdec_av1_slice_work_buffer {
+ struct vdec_av1_slice_mem mv_addr;
+ struct vdec_av1_slice_mem cdf_addr;
+ struct vdec_av1_slice_mem segid_addr;
+};
+
+/**
+ * struct vdec_av1_slice_frame_info - frame info for each slot
+ * @frame_type: frame type
+ * @frame_is_intra: is intra frame
+ * @order_hint: order hint
+ * @order_hints: referece frame order hint
+ * @upscaled_width: upscale width
+ * @pic_pitch: buffer pitch
+ * @frame_width: frane width
+ * @frame_height: frame height
+ * @mi_rows: rows in mode info
+ * @mi_cols: cols in mode info
+ * @ref_count: mark to reference frame counts
+ */
+struct vdec_av1_slice_frame_info {
+ u8 frame_type;
+ u8 frame_is_intra;
+ int order_hint;
+ u32 order_hints[V4L2_AV1_REFS_PER_FRAME];
+ u32 upscaled_width;
+ u32 pic_pitch;
+ u32 frame_width;
+ u32 frame_height;
+ u32 mi_rows;
+ u32 mi_cols;
+ int ref_count;
+};
+
+/**
+ * struct vdec_av1_slice_slot - slot info that needs to be saved in the global instance
+ * @frame_info: frame info for each slot
+ * @timestamp: time stamp info
+ */
+struct vdec_av1_slice_slot {
+ struct vdec_av1_slice_frame_info frame_info[AV1_MAX_FRAME_BUF_COUNT];
+ u64 timestamp[AV1_MAX_FRAME_BUF_COUNT];
+};
+
+/**
+ * struct vdec_av1_slice_fb - frame buffer for decoding
+ * @y: current y buffer address info
+ * @c: current c buffer address info
+ */
+struct vdec_av1_slice_fb {
+ struct vdec_av1_slice_mem y;
+ struct vdec_av1_slice_mem c;
+};
+
+/**
+ * struct vdec_av1_slice_vsi - exchange frame information between Main CPU and MicroP
+ * @bs: input buffer info
+ * @work_buffer: working buffe for hw
+ * @cdf_table: cdf_table buffer
+ * @cdf_tmp: cdf temp buffer
+ * @rd_mv: mv buffer for lat output , core input
+ * @ube: ube buffer
+ * @trans: transcoded buffer
+ * @err_map: err map buffer
+ * @row_info: row info buffer
+ * @fb: current y/c buffer
+ * @ref: ref y/c buffer
+ * @iq_table: iq table buffer
+ * @tile: tile buffer
+ * @slots: slots info for each frame
+ * @slot_id: current frame slot id
+ * @frame: current frame info
+ * @state: status after decode done
+ * @cur_lst_tile_id: tile id for large scale
+ */
+struct vdec_av1_slice_vsi {
+ /* lat */
+ struct vdec_av1_slice_mem bs;
+ struct vdec_av1_slice_work_buffer work_buffer[AV1_MAX_FRAME_BUF_COUNT];
+ struct vdec_av1_slice_mem cdf_table;
+ struct vdec_av1_slice_mem cdf_tmp;
+ /* LAT stage's output, Core stage's input */
+ struct vdec_av1_slice_mem rd_mv;
+ struct vdec_av1_slice_mem ube;
+ struct vdec_av1_slice_mem trans;
+ struct vdec_av1_slice_mem err_map;
+ struct vdec_av1_slice_mem row_info;
+ /* core */
+ struct vdec_av1_slice_fb fb;
+ struct vdec_av1_slice_fb ref[V4L2_AV1_REFS_PER_FRAME];
+ struct vdec_av1_slice_mem iq_table;
+ /* lat and core share*/
+ struct vdec_av1_slice_mem tile;
+ struct vdec_av1_slice_slot slots;
+ u8 slot_id;
+ struct vdec_av1_slice_frame frame;
+ struct vdec_av1_slice_state state;
+ u32 cur_lst_tile_id;
+};
+
+/**
+ * struct vdec_av1_slice_pfc - per-frame context that contains a local vsi.
+ * pass it from lat to core
+ * @vsi: local vsi. copy to/from remote vsi before/after decoding
+ * @ref_idx: reference buffer timestamp
+ * @seq: picture sequence
+ */
+struct vdec_av1_slice_pfc {
+ struct vdec_av1_slice_vsi vsi;
+ u64 ref_idx[V4L2_AV1_REFS_PER_FRAME];
+ int seq;
+};
+
+/**
+ * struct vdec_av1_slice_instance - represent one av1 instance
+ * @ctx: pointer to codec's context
+ * @vpu: VPU instance
+ * @iq_table: iq table buffer
+ * @cdf_table: cdf table buffer
+ * @mv: mv working buffer
+ * @cdf: cdf working buffer
+ * @seg: segmentation working buffer
+ * @cdf_temp: cdf temp buffer
+ * @tile: tile buffer
+ * @slots: slots info
+ * @tile_group: tile_group entry
+ * @level: level of current resolution
+ * @width: width of last picture
+ * @height: height of last picture
+ * @frame_type: frame_type of last picture
+ * @irq: irq to Main CPU or MicroP
+ * @inneracing_mode: is inneracing mode
+ * @init_vsi: vsi used for initialized AV1 instance
+ * @vsi: vsi used for decoding/flush ...
+ * @core_vsi: vsi used for Core stage
+ * @seq: global picture sequence
+ */
+struct vdec_av1_slice_instance {
+ struct mtk_vcodec_ctx *ctx;
+ struct vdec_vpu_inst vpu;
+
+ struct mtk_vcodec_mem iq_table;
+ struct mtk_vcodec_mem cdf_table;
+
+ struct mtk_vcodec_mem mv[AV1_MAX_FRAME_BUF_COUNT];
+ struct mtk_vcodec_mem cdf[AV1_MAX_FRAME_BUF_COUNT];
+ struct mtk_vcodec_mem seg[AV1_MAX_FRAME_BUF_COUNT];
+ struct mtk_vcodec_mem cdf_temp;
+ struct mtk_vcodec_mem tile;
+ struct vdec_av1_slice_slot slots;
+ struct vdec_av1_slice_tile_group tile_group;
+
+ /* for resolution change and get_pic_info */
+ enum vdec_av1_slice_resolution_level level;
+ u32 width;
+ u32 height;
+
+ u32 frame_type;
+ u32 irq;
+ u32 inneracing_mode;
+
+ /* MicroP vsi */
+ union {
+ struct vdec_av1_slice_init_vsi *init_vsi;
+ struct vdec_av1_slice_vsi *vsi;
+ };
+ struct vdec_av1_slice_vsi *core_vsi;
+ int seq;
+};
+
+static int vdec_av1_slice_core_decode(struct vdec_lat_buf *lat_buf);
+
+static inline int vdec_av1_slice_get_msb(u32 n)
+{
+ if (n == 0)
+ return 0;
+ return 31 ^ __builtin_clz(n);
+}
+
+static inline bool vdec_av1_slice_need_scale(u32 ref_width, u32 ref_height,
+ u32 this_width, u32 this_height)
+{
+ return ((this_width << 1) >= ref_width) &&
+ ((this_height << 1) >= ref_height) &&
+ (this_width <= (ref_width << 4)) &&
+ (this_height <= (ref_height << 4));
+}
+
+static void *vdec_av1_get_ctrl_ptr(struct mtk_vcodec_ctx *ctx, int id)
+{
+ struct v4l2_ctrl *ctrl = v4l2_ctrl_find(&ctx->ctrl_hdl, id);
+
+ if (!ctrl)
+ return ERR_PTR(-EINVAL);
+
+ return ctrl->p_cur.p;
+}
I see we keep repeating this kind of a v4l2_ctrl_find() wrapper in drivers.
The only reason this code cannot be factored out is the "context" struct pointer
pointing at structs of different types. Maybe we could
#define v4l2_get_ctrl_ptr(ctx, member, id) \
__v4l2_get_ctrl_ptr((ctx), offsetof(typeof(*ctx), (member)), (id))
void *__v4l2_get_ctrl_ptr(void *ctx, size_t offset, u32 id)
{
struct v4l2_ctrl_handler *hdl = (struct v4l2_ctrl_handler *)(ctx + offset);
struct v4l2_ctrl *ctrl = v4l2_ctrl_find(hdl, id);
if (!ctrl)
return ERR_PTR(-EINVAL);
return ctrl->p_cur.p;
}
and reuse v4l2_get_ctrl_ptr() in drivers?
A similar kind of void* arithmetic happens in container_of, only with '-'.
+
+static int vdec_av1_slice_init_cdf_table(struct vdec_av1_slice_instance *instance)
+{
+ u8 *remote_cdf_table;
+ struct mtk_vcodec_ctx *ctx;
+ struct vdec_av1_slice_init_vsi *vsi;
+ int ret;
+
+ ctx = instance->ctx;
+ vsi = instance->vpu.vsi;
+ if (!ctx || !vsi) {
+ mtk_vcodec_err(instance, "invalid ctx or vsi 0x%p 0x%p\n",
+ ctx, vsi);
+ return -EINVAL;
+ }
The above if block is redundant, because - given the current shape of ths driver
code - the condition is never true.
This function is only called from vdec_av1_slice_init(), where both
instance->ctx and instance->vpu.vsi are set to some values. The latter is even
checked for being null before this function is called.
In the caller, instance->ctx is set to whatever the caller receives from its
caller. This perhaps might be checked, but vdec_av1_slice_init() dereferences
ctx without checking anyway (instance->vpu.codec_type = ctx->current_codec;).
So maybe add a check in vdec_av1_slice_init(), or ensure that
vdec_av1_slice_init() is never passed a NULL ctx.
+
+ remote_cdf_table = mtk_vcodec_fw_map_dm_addr(ctx->dev->fw_handler,
+ (u32)vsi->cdf_table_addr);
+ if (IS_ERR(remote_cdf_table)) {
+ mtk_vcodec_err(instance, "failed to map cdf table\n");
+ return PTR_ERR(remote_cdf_table);
+ }
+
+ mtk_vcodec_debug(instance, "map cdf table to 0x%p\n",
+ remote_cdf_table);
+
+ if (instance->cdf_table.va)
+ mtk_vcodec_mem_free(ctx, &instance->cdf_table);
+ instance->cdf_table.size = vsi->cdf_table_size;
+
+ ret = mtk_vcodec_mem_alloc(ctx, &instance->cdf_table);
+ if (ret)
+ return ret;
+
+ memcpy(instance->cdf_table.va, remote_cdf_table, vsi->cdf_table_size);
+
+ return 0;
+}
+
+static int vdec_av1_slice_init_iq_table(struct vdec_av1_slice_instance *instance)
+{
+ u8 *remote_iq_table;
+ struct mtk_vcodec_ctx *ctx;
+ struct vdec_av1_slice_init_vsi *vsi;
+ int ret;
+
+ ctx = instance->ctx;
+ vsi = instance->vpu.vsi;
+ if (!ctx || !vsi) {
+ mtk_vcodec_err(instance, "invalid ctx or vsi 0x%p 0x%p\n",
+ ctx, vsi);
+ return -EINVAL;
+ }
ditto
+
+ remote_iq_table = mtk_vcodec_fw_map_dm_addr(ctx->dev->fw_handler,
+ (u32)vsi->iq_table_addr);
+ if (IS_ERR(remote_iq_table)) {
+ mtk_vcodec_err(instance, "failed to map iq table\n");
+ return PTR_ERR(remote_iq_table);
+ }
+
+ mtk_vcodec_debug(instance, "map iq table to 0x%p\n", remote_iq_table);
+
+ if (instance->iq_table.va)
+ mtk_vcodec_mem_free(ctx, &instance->iq_table);
+ instance->iq_table.size = vsi->iq_table_size;
+
+ ret = mtk_vcodec_mem_alloc(ctx, &instance->iq_table);
+ if (ret)
+ return ret;
+
+ memcpy(instance->iq_table.va, remote_iq_table, vsi->iq_table_size);
+
+ return 0;
+}
+
+static int vdec_av1_slice_get_new_slot(struct vdec_av1_slice_vsi *vsi)
+{
+ struct vdec_av1_slice_slot *slots = &vsi->slots;
+ int new_slot_idx = AV1_INVALID_IDX;
+ int i;
+
+ for (i = 0; i < AV1_MAX_FRAME_BUF_COUNT; i++) {
+ if (slots->frame_info[i].ref_count == 0) {
+ new_slot_idx = i;
+ break;
+ }
+ }
+
+ if (new_slot_idx != AV1_INVALID_IDX) {
+ slots->frame_info[new_slot_idx].ref_count++;
+ slots->timestamp[new_slot_idx] = vsi->frame.cur_ts;
+ }
+
+ return new_slot_idx;
+}
+
+static void vdec_av1_slice_clear_fb(struct vdec_av1_slice_frame_info *frame_info)
static inline void?
+{
+ memset((void *)frame_info, 0, sizeof(struct vdec_av1_slice_frame_info));
+}
+
+static void vdec_av1_slice_decrease_ref_count(struct vdec_av1_slice_slot *slots, int fb_idx)
+{
+ struct vdec_av1_slice_frame_info *frame_info = slots->frame_info;
+
+ if (fb_idx < 0 || fb_idx >= AV1_MAX_FRAME_BUF_COUNT) {
+ mtk_v4l2_err("av1_error: %s() invalid fb_idx %d\n", __func__, fb_idx);
+ return;
+ }
The above if block is redundant, because - given the current shape of this
driver code - the condition is never true.
This function is only called from the below vdec_av1_slice_cleanup_slots().
The fb_idx formal param comes from the caller's slot_id local variable, whose
value is only assigned in the for loop, iterating from 0 to
AV1_MAX_FRAME_BUF_COUNT - 1, inclusive. Hence slot_id is never < 0
nor >= AV1_MAX_FRAME_BUF_COUNT.
+
+ frame_info[fb_idx].ref_count--;
+ if (frame_info[fb_idx].ref_count < 0) {
+ frame_info[fb_idx].ref_count = 0;
+ mtk_v4l2_err("av1_error: %s() fb_idx %d decrease ref_count error\n",
+ __func__, fb_idx);
+ }
+ vdec_av1_slice_clear_fb(&frame_info[fb_idx]);
+}
+
+static void vdec_av1_slice_cleanup_slots(struct vdec_av1_slice_slot *slots,
+ struct vdec_av1_slice_frame *frame,
+ struct v4l2_ctrl_av1_frame *ctrl_fh)
+{
+ int slot_id, ref_id;
+
+ for (ref_id = 0; ref_id < V4L2_AV1_TOTAL_REFS_PER_FRAME; ref_id++)
+ frame->ref_frame_map[ref_id] = AV1_INVALID_IDX;
+
+ for (slot_id = 0; slot_id < AV1_MAX_FRAME_BUF_COUNT; slot_id++) {
+ u64 timestamp = slots->timestamp[slot_id];
+ bool ref_used = false;
+
+ /* ignored unused slots */
+ if (slots->frame_info[slot_id].ref_count == 0)
+ continue;
+
+ for (ref_id = 0; ref_id < V4L2_AV1_TOTAL_REFS_PER_FRAME; ref_id++) {
+ if (ctrl_fh->reference_frame_ts[ref_id] == timestamp) {
+ frame->ref_frame_map[ref_id] = slot_id;
+ ref_used = true;
+ }
+ }
+
+ if (!ref_used)
+ vdec_av1_slice_decrease_ref_count(slots, slot_id);
+ }
+}
+
+static void vdec_av1_slice_setup_slot(struct vdec_av1_slice_instance *instance,
+ struct vdec_av1_slice_vsi *vsi,
+ struct v4l2_ctrl_av1_frame *ctrl_fh)
+{
+ struct vdec_av1_slice_frame_info *cur_frame_info;
+ struct vdec_av1_slice_uncompressed_header *uh = &vsi->frame.uh;
+ int ref_id;
+
+ memcpy(&vsi->slots, &instance->slots, sizeof(instance->slots));
+ vdec_av1_slice_cleanup_slots(&vsi->slots, &vsi->frame, ctrl_fh);
+ vsi->slot_id = vdec_av1_slice_get_new_slot(vsi);
+
+ if (vsi->slot_id == AV1_INVALID_IDX) {
+ mtk_v4l2_err("warning:av1 get invalid index slot\n");
+ vsi->slot_id = 0;
+ }
+ cur_frame_info = &vsi->slots.frame_info[vsi->slot_id];
+ cur_frame_info->frame_type = uh->frame_type;
+ cur_frame_info->frame_is_intra = ((uh->frame_type == AV1_INTRA_ONLY_FRAME) ||
+ (uh->frame_type == AV1_KEY_FRAME));
+ cur_frame_info->order_hint = uh->order_hint;
+ cur_frame_info->upscaled_width = uh->upscaled_width;
+ cur_frame_info->pic_pitch = 0;
+ cur_frame_info->frame_width = uh->frame_width;
+ cur_frame_info->frame_height = uh->frame_height;
+ cur_frame_info->mi_cols = ((uh->frame_width + 7) >> 3) << 1;
+ cur_frame_info->mi_rows = ((uh->frame_height + 7) >> 3) << 1;
+
+ /* ensure current frame is properly mapped if referenced */
+ for (ref_id = 0; ref_id < V4L2_AV1_TOTAL_REFS_PER_FRAME; ref_id++) {
+ u64 timestamp = vsi->slots.timestamp[vsi->slot_id];
+
+ if (ctrl_fh->reference_frame_ts[ref_id] == timestamp)
+ vsi->frame.ref_frame_map[ref_id] = vsi->slot_id;
+ }
+}
+
+static int vdec_av1_slice_alloc_working_buffer(struct vdec_av1_slice_instance *instance,
+ struct vdec_av1_slice_vsi *vsi)
+{
+ struct mtk_vcodec_ctx *ctx = instance->ctx;
+ struct vdec_av1_slice_work_buffer *work_buffer = vsi->work_buffer;
+ enum vdec_av1_slice_resolution_level level;
+ u32 max_sb_w, max_sb_h, max_w, max_h, w, h;
+ size_t size;
+ int i, ret;
+
+ w = vsi->frame.uh.frame_width;
+ h = vsi->frame.uh.frame_height;
+
+ if (w > VCODEC_DEC_4K_CODED_WIDTH || h > VCODEC_DEC_4K_CODED_HEIGHT)
+ /* 8K */
+ return -EINVAL;
+
+ if (w > MTK_VDEC_MAX_W || h > MTK_VDEC_MAX_H) {
+ /* 4K */
+ level = AV1_RES_4K;
+ max_w = VCODEC_DEC_4K_CODED_WIDTH;
+ max_h = VCODEC_DEC_4K_CODED_HEIGHT;
+ } else {
+ /* FHD */
+ level = AV1_RES_FHD;
+ max_w = MTK_VDEC_MAX_W;
+ max_h = MTK_VDEC_MAX_H;
+ }
+
+ if (level == instance->level)
+ return 0;
+
+ mtk_vcodec_debug(instance, "resolution level changed from %u to %u, %ux%u",
+ instance->level, level, w, h);
+
+ max_sb_w = DIV_ROUND_UP(max_w, 128);
+ max_sb_h = DIV_ROUND_UP(max_h, 128);
+ size = max_sb_w * max_sb_h * SZ_1K;
+ for (i = 0; i < AV1_MAX_FRAME_BUF_COUNT; i++) {
+ if (instance->mv[i].va)
+ mtk_vcodec_mem_free(ctx, &instance->mv[i]);
+ instance->mv[i].size = size;
+ ret = mtk_vcodec_mem_alloc(ctx, &instance->mv[i]);
+ if (ret)
+ goto err;
Please ignore this comment if this has been discussed and settled.
Maybe it's just me, but I feel it is idiomatic in the kernel to
undo all previous allocations if at some iteration we fail. Here a different
approach is taken: we stop iterating and return an error, and free next time
we are called. Why?
+ work_buffer[i].mv_addr.buf = instance->mv[i].dma_addr;
+ work_buffer[i].mv_addr.size = size; > + }
+
+ size = max_sb_w * max_sb_h * 512;
+ for (i = 0; i < AV1_MAX_FRAME_BUF_COUNT; i++) {
+ if (instance->seg[i].va)
+ mtk_vcodec_mem_free(ctx, &instance->seg[i]);
+ instance->seg[i].size = size;
+ ret = mtk_vcodec_mem_alloc(ctx, &instance->seg[i]);
+ if (ret)
+ goto err;
+ work_buffer[i].segid_addr.buf = instance->seg[i].dma_addr;
+ work_buffer[i].segid_addr.size = size;
+ }
+
+ size = 16384;
#define a named constant for this magic number?
+ for (i = 0; i < AV1_MAX_FRAME_BUF_COUNT; i++) {
+ if (instance->cdf[i].va)
+ mtk_vcodec_mem_free(ctx, &instance->cdf[i]);
+ instance->cdf[i].size = size;
+ ret = mtk_vcodec_mem_alloc(ctx, &instance->cdf[i]);
+ if (ret)
+ goto err;
+ work_buffer[i].cdf_addr.buf = instance->cdf[i].dma_addr;
+ work_buffer[i].cdf_addr.size = size;
+ }
The 3 for loops are supposed to iterate from 0 to AV1_MAX_FRAME_BUF_COUNT - 1,
inclusive. Is it possible to merge them?
+ if (!instance->cdf_temp.va) {
+ instance->cdf_temp.size = (SZ_1K * 16 * 100);
+ ret = mtk_vcodec_mem_alloc(ctx, &instance->cdf_temp);
+ if (ret)
+ goto err;
+ vsi->cdf_tmp.buf = instance->cdf_temp.dma_addr;
+ vsi->cdf_tmp.size = instance->cdf_temp.size;
+ }
+ size = AV1_TILE_BUF_SIZE * V4L2_AV1_MAX_TILE_COUNT;
This "size" is never changed until the end of this function.
It is a compile-time constant, so there's no need to assign its
value to an intermediate variable.
+
+ if (instance->tile.va)
+ mtk_vcodec_mem_free(ctx, &instance->tile);
+ instance->tile.size = size;
instance->tile.size = AV1_TILE_BUF_SIZE * V4L2_AV1_MAX_TILE_COUNT;
+
+ ret = mtk_vcodec_mem_alloc(ctx, &instance->tile);
+ if (ret)
+ goto err;
+
+ vsi->tile.buf = instance->tile.dma_addr;
+ vsi->tile.size = size;
vsi->tile.size = instance->tile.size;
and now it is clear the size in vsi is the same as the one in instance.
BTW, is vsi->tile.size supposed to always be equal to the one in instance?
If yes:
- Is instance available whenever we need to access vsi->tile.size?
- What's the point of duplicating this value? Can it be stored in one place?
+
+ instance->level = level;
+ return 0;
+
+err:
+ instance->level = AV1_RES_NONE;
+ return ret;
+}
+
+static void vdec_av1_slice_free_working_buffer(struct vdec_av1_slice_instance *instance)
+{
+ struct mtk_vcodec_ctx *ctx = instance->ctx;
+ int i;
+
+ for (i = 0; i < ARRAY_SIZE(instance->mv); i++)
+ if (instance->mv[i].va)
+ mtk_vcodec_mem_free(ctx, &instance->mv[i]);
Perhaps mtk_vcodec_mem_free() can properly handle the case
(!instance->mv[i].va) ? This would eliminate 7 of 20 lines of code
in this function.
+
+ for (i = 0; i < ARRAY_SIZE(instance->seg); i++)
+ if (instance->seg[i].va)
+ mtk_vcodec_mem_free(ctx, &instance->seg[i]);
+
+ for (i = 0; i < ARRAY_SIZE(instance->cdf); i++)
+ if (instance->cdf[i].va)
+ mtk_vcodec_mem_free(ctx, &instance->cdf[i]);
+
+ if (instance->tile.va)
+ mtk_vcodec_mem_free(ctx, &instance->tile);
+ if (instance->cdf_temp.va)
+ mtk_vcodec_mem_free(ctx, &instance->cdf_temp);
+ if (instance->cdf_table.va)
+ mtk_vcodec_mem_free(ctx, &instance->cdf_table);
+ if (instance->iq_table.va)
+ mtk_vcodec_mem_free(ctx, &instance->iq_table);
+
+ instance->level = AV1_RES_NONE;
+}
+
+static void vdec_av1_slice_vsi_from_remote(struct vdec_av1_slice_vsi *vsi,
+ struct vdec_av1_slice_vsi *remote_vsi)
static inline void?
+{
+ memcpy(&vsi->trans, &remote_vsi->trans, sizeof(vsi->trans));
+ memcpy(&vsi->state, &remote_vsi->state, sizeof(vsi->state));
+}
+
+static void vdec_av1_slice_vsi_to_remote(struct vdec_av1_slice_vsi *vsi,
+ struct vdec_av1_slice_vsi *remote_vsi)
static inline void?
+{
+ memcpy(remote_vsi, vsi, sizeof(*vsi));
+}
+
+static int vdec_av1_slice_setup_lat_from_src_buf(struct vdec_av1_slice_instance *instance,
+ struct vdec_av1_slice_vsi *vsi,
+ struct vdec_lat_buf *lat_buf)
+{
+ struct vb2_v4l2_buffer *src;
+ struct vb2_v4l2_buffer *dst;
+
+ src = v4l2_m2m_next_src_buf(instance->ctx->m2m_ctx);
+ if (!src)
+ return -EINVAL;
+
+ lat_buf->src_buf_req = src->vb2_buf.req_obj.req;
+ dst = &lat_buf->ts_info;
the "ts_info" actually contains a struct vb2_v4l2_buffer. Why such a name?
+ v4l2_m2m_buf_copy_metadata(src, dst, true);
+ vsi->frame.cur_ts = dst->vb2_buf.timestamp;
+
+ return 0;
+}
+
+static short vdec_av1_slice_resolve_divisor_32(u32 D, short *shift)
+{
+ int f;
+ int e;
+
+ *shift = vdec_av1_slice_get_msb(D);
+ /* e is obtained from D after resetting the most significant 1 bit. */
+ e = D - ((u32)1 << *shift);
+ /* Get the most significant DIV_LUT_BITS (8) bits of e into f */
+ if (*shift > DIV_LUT_BITS)
+ f = AV1_DIV_ROUND_UP_POW2(e, *shift - DIV_LUT_BITS);
+ else
+ f = e << (DIV_LUT_BITS - *shift);
+ if (f > DIV_LUT_NUM)
+ return -1;
+ *shift += DIV_LUT_PREC_BITS;
+ /* Use f as lookup into the precomputed table of multipliers */
+ return div_lut[f];
+}
+
+static void vdec_av1_slice_get_shear_params(struct vdec_av1_slice_gm *gm_params)
+{
+ const int *mat = gm_params->wmmat;
+ short shift;
+ short y;
+ long long gv, dv;
+
+ if (gm_params->wmmat[2] <= 0)
+ return;
+
+ gm_params->alpha = clamp_val(mat[2] - (1 << WARPEDMODEL_PREC_BITS), S16_MIN, S16_MAX);
+ gm_params->beta = clamp_val(mat[3], S16_MIN, S16_MAX);
+
+ y = vdec_av1_slice_resolve_divisor_32(abs(mat[2]), &shift) * (mat[2] < 0 ? -1 : 1);
+
+ gv = ((long long)mat[4] * (1 << WARPEDMODEL_PREC_BITS)) * y;
+ gm_params->gamma = clamp_val((int)AV1_DIV_ROUND_UP_POW2_SIGNED(gv, shift),
+ S16_MIN, S16_MAX);
+
+ dv = ((long long)mat[3] * mat[4]) * y;
+ gm_params->delta = clamp_val(mat[5] - (int)AV1_DIV_ROUND_UP_POW2_SIGNED(dv, shift) -
+ (1 << WARPEDMODEL_PREC_BITS), S16_MIN, S16_MAX);
+
+ gm_params->alpha = AV1_DIV_ROUND_UP_POW2_SIGNED(gm_params->alpha, WARP_PARAM_REDUCE_BITS) *
+ (1 << WARP_PARAM_REDUCE_BITS);
+ gm_params->beta = AV1_DIV_ROUND_UP_POW2_SIGNED(gm_params->beta, WARP_PARAM_REDUCE_BITS) *
+ (1 << WARP_PARAM_REDUCE_BITS);
+ gm_params->gamma = AV1_DIV_ROUND_UP_POW2_SIGNED(gm_params->gamma, WARP_PARAM_REDUCE_BITS) *
+ (1 << WARP_PARAM_REDUCE_BITS);
+ gm_params->delta = AV1_DIV_ROUND_UP_POW2_SIGNED(gm_params->delta, WARP_PARAM_REDUCE_BITS) *
+ (1 << WARP_PARAM_REDUCE_BITS);
+}
+
+static void vdec_av1_slice_setup_gm(struct vdec_av1_slice_gm *gm,
+ struct v4l2_av1_global_motion *ctrl_gm)
+{
+ u32 i, j;
+
+ for (i = 0; i < V4L2_AV1_TOTAL_REFS_PER_FRAME; i++) {
+ gm[i].wmtype = ctrl_gm->type[i];
+ for (j = 0; j < 6; j++)
Maybe #define this magic 6?
+ gm[i].wmmat[j] = ctrl_gm->params[i][j];
+
+ gm[i].invalid = !!(ctrl_gm->invalid & BIT(i));
+ gm[i].alpha = 0;
+ gm[i].beta = 0;
+ gm[i].gamma = 0;
+ gm[i].delta = 0;
+ if (gm[i].wmtype <= 3)
And this 3?
+ vdec_av1_slice_get_shear_params(&gm[i]);
+ }
+}
+
+static void vdec_av1_slice_setup_seg(struct vdec_av1_slice_seg *seg,
+ struct v4l2_av1_segmentation *ctrl_seg)
+{
+ u32 i, j;
+
+ seg->segmentation_enabled = SEGMENTATION_FLAG(ctrl_seg, ENABLED);
+ seg->segmentation_update_map = SEGMENTATION_FLAG(ctrl_seg, UPDATE_MAP);
+ seg->segmentation_temporal_update = SEGMENTATION_FLAG(ctrl_seg, TEMPORAL_UPDATE);
+ seg->segmentation_update_data = SEGMENTATION_FLAG(ctrl_seg, UPDATE_DATA);
+ seg->segid_preskip = SEGMENTATION_FLAG(ctrl_seg, SEG_ID_PRE_SKIP);
+ seg->last_active_segid = ctrl_seg->last_active_seg_id;
+
+ for (i = 0; i < V4L2_AV1_MAX_SEGMENTS; i++) {
+ seg->feature_enabled_mask[i] = ctrl_seg->feature_enabled[i];
+ for (j = 0; j < V4L2_AV1_SEG_LVL_MAX; j++)
+ seg->feature_data[i][j] = ctrl_seg->feature_data[i][j];
+ }
+}
+
+static void vdec_av1_slice_setup_quant(struct vdec_av1_slice_quantization *quant,
+ struct v4l2_av1_quantization *ctrl_quant)
+{
+ quant->base_q_idx = ctrl_quant->base_q_idx;
+ quant->delta_qydc = ctrl_quant->delta_q_y_dc;
+ quant->delta_qudc = ctrl_quant->delta_q_u_dc;
+ quant->delta_quac = ctrl_quant->delta_q_u_ac;
+ quant->delta_qvdc = ctrl_quant->delta_q_v_dc;
+ quant->delta_qvac = ctrl_quant->delta_q_v_ac;
+ quant->qm_y = ctrl_quant->qm_y;
+ quant->qm_u = ctrl_quant->qm_u;
+ quant->qm_v = ctrl_quant->qm_v;
Can a common struct be introduced to hold these parameters?
And then copied in one go?
Maybe there's a good reason the code is the way it is now. However,
a series of "dumb" assignments (no value modifications) makes me wonder.
+ quant->using_qmatrix = QUANT_FLAG(ctrl_quant, USING_QMATRIX);
+}
+
+static int vdec_av1_slice_get_qindex(struct vdec_av1_slice_uncompressed_header *uh,
+ int segmentation_id)
+{
+ struct vdec_av1_slice_seg *seg = &uh->seg;
+ struct vdec_av1_slice_quantization *quant = &uh->quant;
+ int data = 0, qindex = 0;
+
+ if (seg->segmentation_enabled &&
+ (seg->feature_enabled_mask[segmentation_id] & BIT(SEG_LVL_ALT_Q))) {
+ data = seg->feature_data[segmentation_id][SEG_LVL_ALT_Q];
+ qindex = quant->base_q_idx + data;
+ return clamp_val(qindex, 0, MAXQ);
+ }
+
+ return quant->base_q_idx;
+}
+
+static void vdec_av1_slice_setup_lr(struct vdec_av1_slice_lr *lr,
+ struct v4l2_av1_loop_restoration *ctrl_lr)
+{
+ int i;
+
+ lr->use_lr = 0;
+ lr->use_chroma_lr = 0;
+ for (i = 0; i < V4L2_AV1_NUM_PLANES_MAX; i++) {
+ lr->frame_restoration_type[i] = ctrl_lr->frame_restoration_type[i];
+ lr->loop_restoration_size[i] = ctrl_lr->loop_restoration_size[i];
+ if (lr->frame_restoration_type[i]) {
+ lr->use_lr = 1;
+ if (i > 0)
+ lr->use_chroma_lr = 1;
+ }
+ }
+}
+
+static void vdec_av1_slice_setup_lf(struct vdec_av1_slice_loop_filter *lf,
+ struct v4l2_av1_loop_filter *ctrl_lf)
+{
+ int i;
+
+ for (i = 0; i < ARRAY_SIZE(lf->loop_filter_level); i++)
+ lf->loop_filter_level[i] = ctrl_lf->level[i];
+
+ for (i = 0; i < V4L2_AV1_TOTAL_REFS_PER_FRAME; i++)
+ lf->loop_filter_ref_deltas[i] = ctrl_lf->ref_deltas[i];
+
+ for (i = 0; i < ARRAY_SIZE(lf->loop_filter_mode_deltas); i++)
+ lf->loop_filter_mode_deltas[i] = ctrl_lf->mode_deltas[i];
+
+ lf->loop_filter_sharpness = ctrl_lf->sharpness;
+ lf->loop_filter_delta_enabled =
+ BIT_FLAG(ctrl_lf, V4L2_AV1_LOOP_FILTER_FLAG_DELTA_ENABLED);
+}
+
+static void vdec_av1_slice_setup_cdef(struct vdec_av1_slice_cdef *cdef,
+ struct v4l2_av1_cdef *ctrl_cdef)
+{
+ int i;
+
+ cdef->cdef_damping = ctrl_cdef->damping_minus_3 + 3;
+ cdef->cdef_bits = ctrl_cdef->bits;
+
+ for (i = 0; i < V4L2_AV1_CDEF_MAX; i++) {
+ if (ctrl_cdef->y_sec_strength[i] == 4)
+ ctrl_cdef->y_sec_strength[i] -= 1;
+
+ if (ctrl_cdef->uv_sec_strength[i] == 4)
+ ctrl_cdef->uv_sec_strength[i] -= 1;
+
+ cdef->cdef_y_strength[i] =
+ ctrl_cdef->y_pri_strength[i] << SECONDARY_FILTER_STRENGTH_NUM_BITS |
+ ctrl_cdef->y_sec_strength[i];
+ cdef->cdef_uv_strength[i] =
+ ctrl_cdef->uv_pri_strength[i] << SECONDARY_FILTER_STRENGTH_NUM_BITS |
+ ctrl_cdef->uv_sec_strength[i];
+ }
+}
Both vdec_av1_slice_setup_lf() and vdec_av1_slice_setup_cdef():
I'm wondering if the user of struct vdec_av1_slice_loop_filter and struct
vdec_av1_slice_cdef could work with the uAPI variants of these structs? Is there
a need for driver-specific mutations? (Maybe there is, the driver's author
should know).
+
+static void vdec_av1_slice_setup_seq(struct vdec_av1_slice_seq_header *seq,
+ struct v4l2_ctrl_av1_sequence *ctrl_seq)
+{
+ seq->bitdepth = ctrl_seq->bit_depth;
+ seq->max_frame_width = ctrl_seq->max_frame_width_minus_1 + 1;
+ seq->max_frame_height = ctrl_seq->max_frame_height_minus_1 + 1;
+ seq->enable_superres = SEQUENCE_FLAG(ctrl_seq, ENABLE_SUPERRES);
+ seq->enable_filter_intra = SEQUENCE_FLAG(ctrl_seq, ENABLE_FILTER_INTRA);
+ seq->enable_intra_edge_filter = SEQUENCE_FLAG(ctrl_seq, ENABLE_INTRA_EDGE_FILTER);
+ seq->enable_interintra_compound = SEQUENCE_FLAG(ctrl_seq, ENABLE_INTERINTRA_COMPOUND);
+ seq->enable_masked_compound = SEQUENCE_FLAG(ctrl_seq, ENABLE_MASKED_COMPOUND);
+ seq->enable_dual_filter = SEQUENCE_FLAG(ctrl_seq, ENABLE_DUAL_FILTER);
+ seq->enable_jnt_comp = SEQUENCE_FLAG(ctrl_seq, ENABLE_JNT_COMP);
+ seq->mono_chrome = SEQUENCE_FLAG(ctrl_seq, MONO_CHROME);
+ seq->enable_order_hint = SEQUENCE_FLAG(ctrl_seq, ENABLE_ORDER_HINT);
+ seq->order_hint_bits = ctrl_seq->order_hint_bits;
+ seq->use_128x128_superblock = SEQUENCE_FLAG(ctrl_seq, USE_128X128_SUPERBLOCK);
+ seq->subsampling_x = SEQUENCE_FLAG(ctrl_seq, SUBSAMPLING_X);
+ seq->subsampling_y = SEQUENCE_FLAG(ctrl_seq, SUBSAMPLING_Y);
+}
+
+static void vdec_av1_slice_setup_tile(struct vdec_av1_slice_frame *frame,
+ struct v4l2_av1_tile_info *ctrl_tile)
+{
+ struct vdec_av1_slice_seq_header *seq = &frame->seq;
+ struct vdec_av1_slice_tile *tile = &frame->uh.tile;
+ u32 mib_size_log2 = seq->use_128x128_superblock ? 5 : 4;
+ int i;
+
+ tile->tile_cols = ctrl_tile->tile_cols;
+ tile->tile_rows = ctrl_tile->tile_rows;
+ tile->context_update_tile_id = ctrl_tile->context_update_tile_id;
+ tile->uniform_tile_spacing_flag =
+ BIT_FLAG(ctrl_tile, V4L2_AV1_TILE_INFO_FLAG_UNIFORM_TILE_SPACING);
+
+ for (i = 0; i < tile->tile_cols + 1; i++)
+ tile->mi_col_starts[i] =
+ ALIGN(ctrl_tile->mi_col_starts[i], BIT(mib_size_log2)) >> mib_size_log2;
+
+ for (i = 0; i < tile->tile_rows + 1; i++)
+ tile->mi_row_starts[i] =
+ ALIGN(ctrl_tile->mi_row_starts[i], BIT(mib_size_log2)) >> mib_size_log2;
+}
+
+static void vdec_av1_slice_setup_uh(struct vdec_av1_slice_instance *instance,
+ struct vdec_av1_slice_frame *frame,
+ struct v4l2_ctrl_av1_frame *ctrl_fh)
+{
+ struct vdec_av1_slice_uncompressed_header *uh = &frame->uh;
+ int i;
+
+ uh->use_ref_frame_mvs = FH_FLAG(ctrl_fh, USE_REF_FRAME_MVS);
+ uh->order_hint = ctrl_fh->order_hint;
+ vdec_av1_slice_setup_gm(uh->gm, &ctrl_fh->global_motion);
+ uh->upscaled_width = ctrl_fh->upscaled_width;
+ uh->frame_width = ctrl_fh->frame_width_minus_1 + 1;
+ uh->frame_height = ctrl_fh->frame_height_minus_1 + 1;
+ uh->mi_cols = ((uh->frame_width + 7) >> 3) << 1;
+ uh->mi_rows = ((uh->frame_height + 7) >> 3) << 1;
+ uh->reduced_tx_set = FH_FLAG(ctrl_fh, REDUCED_TX_SET);
+ uh->tx_mode = ctrl_fh->tx_mode;
+ uh->uniform_tile_spacing_flag = FH_FLAG(ctrl_fh, UNIFORM_TILE_SPACING);
+ uh->interpolation_filter = ctrl_fh->interpolation_filter;
+ uh->allow_warped_motion = FH_FLAG(ctrl_fh, ALLOW_WARPED_MOTION);
+ uh->is_motion_mode_switchable = FH_FLAG(ctrl_fh, IS_MOTION_MODE_SWITCHABLE);
+ uh->frame_type = ctrl_fh->frame_type;
+ uh->frame_is_intra = (uh->frame_type == V4L2_AV1_INTRA_ONLY_FRAME ||
+ uh->frame_type == V4L2_AV1_KEY_FRAME);
+
+ if (!uh->frame_is_intra && FH_FLAG(ctrl_fh, REFERENCE_SELECT))
+ uh->reference_mode = AV1_REFERENCE_MODE_SELECT;
+ else
+ uh->reference_mode = AV1_SINGLE_REFERENCE;
+
+ uh->allow_high_precision_mv = FH_FLAG(ctrl_fh, ALLOW_HIGH_PRECISION_MV);
+ uh->allow_intra_bc = FH_FLAG(ctrl_fh, ALLOW_INTRABC);
+ uh->force_integer_mv = FH_FLAG(ctrl_fh, FORCE_INTEGER_MV);
+ uh->allow_screen_content_tools = FH_FLAG(ctrl_fh, ALLOW_SCREEN_CONTENT_TOOLS);
+ uh->error_resilient_mode = FH_FLAG(ctrl_fh, ERROR_RESILIENT_MODE);
+ uh->primary_ref_frame = ctrl_fh->primary_ref_frame;
+ uh->disable_frame_end_update_cdf =
+ FH_FLAG(ctrl_fh, DISABLE_FRAME_END_UPDATE_CDF);
+ uh->disable_cdf_update = FH_FLAG(ctrl_fh, DISABLE_CDF_UPDATE);
+ uh->skip_mode.skip_mode_present = FH_FLAG(ctrl_fh, SKIP_MODE_PRESENT);
+ uh->skip_mode.skip_mode_frame[0] =
+ ctrl_fh->skip_mode_frame[0] - V4L2_AV1_REF_LAST_FRAME;
+ uh->skip_mode.skip_mode_frame[1] =
+ ctrl_fh->skip_mode_frame[1] - V4L2_AV1_REF_LAST_FRAME;
+ uh->skip_mode.skip_mode_allowed = ctrl_fh->skip_mode_frame[0] ? 1 : 0;
+
+ vdec_av1_slice_setup_seg(&uh->seg, &ctrl_fh->segmentation);
+ uh->delta_q_lf.delta_q_present = QUANT_FLAG(&ctrl_fh->quantization, DELTA_Q_PRESENT);
+ uh->delta_q_lf.delta_q_res = 1 << ctrl_fh->quantization.delta_q_res;
+ uh->delta_q_lf.delta_lf_present =
+ BIT_FLAG(&ctrl_fh->loop_filter, V4L2_AV1_LOOP_FILTER_FLAG_DELTA_LF_PRESENT);
+ uh->delta_q_lf.delta_lf_res = ctrl_fh->loop_filter.delta_lf_res;
+ uh->delta_q_lf.delta_lf_multi =
+ BIT_FLAG(&ctrl_fh->loop_filter, V4L2_AV1_LOOP_FILTER_FLAG_DELTA_LF_MULTI);
+ vdec_av1_slice_setup_quant(&uh->quant, &ctrl_fh->quantization);
+
+ uh->coded_loss_less = 1;
+ for (i = 0; i < V4L2_AV1_MAX_SEGMENTS; i++) {
+ uh->quant.qindex[i] = vdec_av1_slice_get_qindex(uh, i);
+ uh->loss_less_array[i] =
+ (uh->quant.qindex[i] == 0 && uh->quant.delta_qydc == 0 &&
+ uh->quant.delta_quac == 0 && uh->quant.delta_qudc == 0 &&
+ uh->quant.delta_qvac == 0 && uh->quant.delta_qvdc == 0);
+
+ if (!uh->loss_less_array[i])
+ uh->coded_loss_less = 0;
+ }
+
+ vdec_av1_slice_setup_lr(&uh->lr, &ctrl_fh->loop_restoration);
+ uh->superres_denom = ctrl_fh->superres_denom;
+ vdec_av1_slice_setup_lf(&uh->loop_filter, &ctrl_fh->loop_filter);
+ vdec_av1_slice_setup_cdef(&uh->cdef, &ctrl_fh->cdef);
+ vdec_av1_slice_setup_tile(frame, &ctrl_fh->tile_info);
+}
+
+static int vdec_av1_slice_setup_tile_group(struct vdec_av1_slice_instance *instance,
+ struct vdec_av1_slice_vsi *vsi)
+{
+ struct v4l2_ctrl_av1_tile_group_entry *ctrl_tge;
+ struct vdec_av1_slice_tile_group *tile_group = &instance->tile_group;
+ struct vdec_av1_slice_uncompressed_header *uh = &vsi->frame.uh;
+ struct vdec_av1_slice_tile *tile = &uh->tile;
+ struct v4l2_ctrl *ctrl;
+ u32 tge_size;
+ int i;
+
+ ctrl = v4l2_ctrl_find(&instance->ctx->ctrl_hdl, V4L2_CID_STATELESS_AV1_TILE_GROUP_ENTRY);
+ if (!ctrl)
+ return -EINVAL;
+
+ tge_size = ctrl->elems;
+ ctrl_tge = (struct v4l2_ctrl_av1_tile_group_entry *)ctrl->p_cur.p;
+
+ tile_group->num_tiles = tile->tile_cols * tile->tile_rows;
+
+ if (tile_group->num_tiles != tge_size ||
+ tile_group->num_tiles > V4L2_AV1_MAX_TILE_COUNT) {
+ mtk_vcodec_err(instance, "invalid tge_size %d, tile_num:%d\n",
+ tge_size, tile_group->num_tiles);
+ return -EINVAL;
+ }
+
+ for (i = 0; i < tge_size; i++) {
+ if (i != ctrl_tge[i].tile_row * vsi->frame.uh.tile.tile_cols +
+ ctrl_tge[i].tile_col) {
+ mtk_vcodec_err(instance, "invalid tge info %d, %d %d %d\n",
+ i, ctrl_tge[i].tile_row, ctrl_tge[i].tile_col,
+ vsi->frame.uh.tile.tile_rows);
+ return -EINVAL;
+ }
+ tile_group->tile_size[i] = ctrl_tge[i].tile_size;
+ tile_group->tile_start_offset[i] = ctrl_tge[i].tile_offset;
+ }
+
+ return 0;
+}
+
+static void vdec_av1_slice_setup_state(struct vdec_av1_slice_vsi *vsi)
static inline void?
+{
+ memset(&vsi->state, 0, sizeof(vsi->state));
+}
+
+static void vdec_av1_slice_setup_scale_factors(struct vdec_av1_slice_frame_refs *frame_ref,
+ struct vdec_av1_slice_frame_info *ref_frame_info,
+ struct vdec_av1_slice_uncompressed_header *uh)
+{
+ struct vdec_av1_slice_scale_factors *scale_factors = &frame_ref->scale_factors;
+ u32 ref_upscaled_width = ref_frame_info->upscaled_width;
+ u32 ref_frame_height = ref_frame_info->frame_height;
+ u32 frame_width = uh->frame_width;
+ u32 frame_height = uh->frame_height;
+
+ if (!vdec_av1_slice_need_scale(ref_upscaled_width, ref_frame_height,
+ frame_width, frame_height)) {
+ scale_factors->x_scale = -1;
+ scale_factors->y_scale = -1;
+ scale_factors->is_scaled = 0;
+ return;
+ }
+
+ scale_factors->x_scale =
+ ((ref_upscaled_width << AV1_REF_SCALE_SHIFT) + (frame_width >> 1)) / frame_width;
+ scale_factors->y_scale =
+ ((ref_frame_height << AV1_REF_SCALE_SHIFT) + (frame_height >> 1)) / frame_height;
+ scale_factors->is_scaled =
+ (scale_factors->x_scale != AV1_REF_INVALID_SCALE) &&
+ (scale_factors->y_scale != AV1_REF_INVALID_SCALE) &&
+ (scale_factors->x_scale != AV1_REF_NO_SCALE ||
+ scale_factors->y_scale != AV1_REF_NO_SCALE);
+ scale_factors->x_step =
+ AV1_DIV_ROUND_UP_POW2(scale_factors->x_scale,
+ AV1_REF_SCALE_SHIFT - AV1_SCALE_SUBPEL_BITS);
+ scale_factors->y_step =
+ AV1_DIV_ROUND_UP_POW2(scale_factors->y_scale,
+ AV1_REF_SCALE_SHIFT - AV1_SCALE_SUBPEL_BITS);
+}
+
+static int vdec_av1_slice_get_relative_dist(int a, int b, u8 enable_order_hint, u8 order_hint_bits)
+{
+ int diff = 0;
+ int m = 0;
+
+ if (!enable_order_hint)
+ return 0;
+
+ diff = a - b;
+ m = 1 << (order_hint_bits - 1);
+ diff = (diff & (m - 1)) - (diff & m);
+
+ return diff;
+}
This function is called in one place only, and its result needs to be
interpreted at call site. Can it return the result in a form expected
at call site...
+
+static void vdec_av1_slice_setup_ref(struct vdec_av1_slice_pfc *pfc,
+ struct v4l2_ctrl_av1_frame *ctrl_fh)
+{
+ struct vdec_av1_slice_vsi *vsi = &pfc->vsi;
+ struct vdec_av1_slice_frame *frame = &vsi->frame;
+ struct vdec_av1_slice_slot *slots = &vsi->slots;
+ struct vdec_av1_slice_uncompressed_header *uh = &frame->uh;
+ struct vdec_av1_slice_seq_header *seq = &frame->seq;
+ struct vdec_av1_slice_frame_info *cur_frame_info =
+ &slots->frame_info[vsi->slot_id];
+ struct vdec_av1_slice_frame_info *frame_info;
+ int i, slot_id;
+
+ if (uh->frame_is_intra)
+ return;
+
+ for (i = 0; i < V4L2_AV1_REFS_PER_FRAME; i++) {
+ int ref_idx = ctrl_fh->ref_frame_idx[i];
+
+ pfc->ref_idx[i] = ctrl_fh->reference_frame_ts[ref_idx];
+ slot_id = frame->ref_frame_map[ref_idx];
+ frame_info = &slots->frame_info[slot_id];
+ if (slot_id == AV1_INVALID_IDX) {
+ mtk_v4l2_err("cannot match reference[%d] 0x%llx\n", i,
+ ctrl_fh->reference_frame_ts[ref_idx]);
+ frame->order_hints[i] = 0;
+ frame->ref_frame_valid[i] = 0;
+ continue;
+ }
+
+ frame->frame_refs[i].ref_fb_idx = slot_id;
+ vdec_av1_slice_setup_scale_factors(&frame->frame_refs[i],
+ frame_info, uh);
+ if (!seq->enable_order_hint)
+ frame->ref_frame_sign_bias[i + 1] = 0;
+ else
+ frame->ref_frame_sign_bias[i + 1] =
+ vdec_av1_slice_get_relative_dist(frame_info->order_hint,
+ uh->order_hint,
+ seq->enable_order_hint,
+ seq->order_hint_bits)
+ <= 0 ? 0 : 1;
... to get rid of this tri-argument operator altogether?
+
+ frame->order_hints[i] = ctrl_fh->order_hints[i + 1];
+ cur_frame_info->order_hints[i] = frame->order_hints[i];
+ frame->ref_frame_valid[i] = 1;
+ }
+}
+
+static void vdec_av1_slice_get_previous(struct vdec_av1_slice_vsi *vsi)
+{
+ struct vdec_av1_slice_frame *frame = &vsi->frame;
+
+ if (frame->uh.primary_ref_frame == 7)
#define magic number 7?
+ frame->prev_fb_idx = AV1_INVALID_IDX;
+ else
+ frame->prev_fb_idx = frame->frame_refs[frame->uh.primary_ref_frame].ref_fb_idx;
+}
+
+static void vdec_av1_slice_setup_operating_mode(struct vdec_av1_slice_instance *instance,
+ struct vdec_av1_slice_frame *frame)
static inline void?
+{
+ frame->large_scale_tile = 0;
+}
+
+static int vdec_av1_slice_setup_pfc(struct vdec_av1_slice_instance *instance,
+ struct vdec_av1_slice_pfc *pfc)
+{
+ struct v4l2_ctrl_av1_frame *ctrl_fh;
+ struct v4l2_ctrl_av1_sequence *ctrl_seq;
+ struct vdec_av1_slice_vsi *vsi = &pfc->vsi;
+ int ret = 0;
+
+ /* frame header */
+ ctrl_fh = (struct v4l2_ctrl_av1_frame *)
+ vdec_av1_get_ctrl_ptr(instance->ctx,
+ V4L2_CID_STATELESS_AV1_FRAME);
+ if (IS_ERR(ctrl_fh))
+ return PTR_ERR(ctrl_fh);
+
+ ctrl_seq = (struct v4l2_ctrl_av1_sequence *)
+ vdec_av1_get_ctrl_ptr(instance->ctx,
+ V4L2_CID_STATELESS_AV1_SEQUENCE);
+ if (IS_ERR(ctrl_seq))
+ return PTR_ERR(ctrl_seq);
Just to make sure: I assume request api is used? If so, does vdec's framework
ensure that v4l2_ctrl_request_setup() has been called? It influences what's
actually in ctrl->p_cur.p (current or previous value), and the
vdec_av1_get_ctrl_ptr() wrapper returns ctrl->p_cur.p.
+
+ /* setup vsi information */
+ vdec_av1_slice_setup_seq(&vsi->frame.seq, ctrl_seq);
+ vdec_av1_slice_setup_uh(instance, &vsi->frame, ctrl_fh);
+ vdec_av1_slice_setup_operating_mode(instance, &vsi->frame);
+
+ vdec_av1_slice_setup_state(vsi);
+ vdec_av1_slice_setup_slot(instance, vsi, ctrl_fh);
+ vdec_av1_slice_setup_ref(pfc, ctrl_fh);
+ vdec_av1_slice_get_previous(vsi);
+
+ pfc->seq = instance->seq;
+ instance->seq++;
+
+ return ret;
+}
+
+static void vdec_av1_slice_setup_lat_buffer(struct vdec_av1_slice_instance *instance,
+ struct vdec_av1_slice_vsi *vsi,
+ struct mtk_vcodec_mem *bs,
+ struct vdec_lat_buf *lat_buf)
+{
+ struct vdec_av1_slice_work_buffer *work_buffer;
+ int i;
+
+ vsi->bs.dma_addr = bs->dma_addr;
+ vsi->bs.size = bs->size;
+
+ vsi->ube.dma_addr = lat_buf->ctx->msg_queue.wdma_addr.dma_addr;
+ vsi->ube.size = lat_buf->ctx->msg_queue.wdma_addr.size;
+ vsi->trans.dma_addr = lat_buf->ctx->msg_queue.wdma_wptr_addr;
+ /* used to store trans end */
+ vsi->trans.dma_addr_end = lat_buf->ctx->msg_queue.wdma_rptr_addr;
+ vsi->err_map.dma_addr = lat_buf->wdma_err_addr.dma_addr;
+ vsi->err_map.size = lat_buf->wdma_err_addr.size;
+ vsi->rd_mv.dma_addr = lat_buf->rd_mv_addr.dma_addr;
+ vsi->rd_mv.size = lat_buf->rd_mv_addr.size;
+
+ vsi->row_info.buf = 0;
+ vsi->row_info.size = 0;
+
+ work_buffer = vsi->work_buffer;
+
+ for (i = 0; i < AV1_MAX_FRAME_BUF_COUNT; i++) {
+ work_buffer[i].mv_addr.buf = instance->mv[i].dma_addr;
+ work_buffer[i].mv_addr.size = instance->mv[i].size;
+ work_buffer[i].segid_addr.buf = instance->seg[i].dma_addr;
+ work_buffer[i].segid_addr.size = instance->seg[i].size;
+ work_buffer[i].cdf_addr.buf = instance->cdf[i].dma_addr;
+ work_buffer[i].cdf_addr.size = instance->cdf[i].size;
+ }
+
+ vsi->cdf_tmp.buf = instance->cdf_temp.dma_addr;
+ vsi->cdf_tmp.size = instance->cdf_temp.size;
+
+ vsi->tile.buf = instance->tile.dma_addr;
+ vsi->tile.size = instance->tile.size;
+ memcpy(lat_buf->tile_addr.va, instance->tile.va, 64 * instance->tile_group.num_tiles);
+
+ vsi->cdf_table.buf = instance->cdf_table.dma_addr;
+ vsi->cdf_table.size = instance->cdf_table.size;
+ vsi->iq_table.buf = instance->iq_table.dma_addr;
+ vsi->iq_table.size = instance->iq_table.size;
+}
+
+static void vdec_av1_slice_setup_seg_buffer(struct vdec_av1_slice_instance *instance,
+ struct vdec_av1_slice_vsi *vsi)
+{
+ struct vdec_av1_slice_uncompressed_header *uh = &vsi->frame.uh;
+ struct mtk_vcodec_mem *buf;
+
+ /* reset segment buffer */
+ if (uh->primary_ref_frame == 7 || !uh->seg.segmentation_enabled) {
#define magic 7?
+ mtk_vcodec_debug(instance, "reset seg %d\n", vsi->slot_id);
+ if (vsi->slot_id != AV1_INVALID_IDX) {
+ buf = &instance->seg[vsi->slot_id];
+ memset(buf->va, 0, buf->size);
+ }
+ }
+}
+
+static void vdec_av1_slice_setup_tile_buffer(struct vdec_av1_slice_instance *instance,
+ struct vdec_av1_slice_vsi *vsi,
+ struct mtk_vcodec_mem *bs)
+{
+ struct vdec_av1_slice_tile_group *tile_group = &instance->tile_group;
+ struct vdec_av1_slice_uncompressed_header *uh = &vsi->frame.uh;
+ struct vdec_av1_slice_tile *tile = &uh->tile;
+ u32 tile_num, tile_row, tile_col;
+ u32 allow_update_cdf = 0;
+ u32 sb_boundary_x_m1 = 0, sb_boundary_y_m1 = 0;
+ int tile_info_base;
+ u32 tile_buf_pa;
+ u32 *tile_info_buf = instance->tile.va;
+ u32 pa = (u32)bs->dma_addr;
+
+ if (uh->disable_cdf_update == 0)
+ allow_update_cdf = 1;
+
+ for (tile_num = 0; tile_num < tile_group->num_tiles; tile_num++) {
+ /* each uint32 takes place of 4 bytes */
+ tile_info_base = (AV1_TILE_BUF_SIZE * tile_num) >> 2;
+ tile_row = tile_num / tile->tile_cols;
+ tile_col = tile_num % tile->tile_cols;
+ tile_info_buf[tile_info_base + 0] = (tile_group->tile_size[tile_num] << 3);
+ tile_buf_pa = pa + tile_group->tile_start_offset[tile_num];
+
+ tile_info_buf[tile_info_base + 1] = (tile_buf_pa >> 4) << 4;
+ tile_info_buf[tile_info_base + 2] = (tile_buf_pa % 16) << 3;
+
+ sb_boundary_x_m1 =
+ (tile->mi_col_starts[tile_col + 1] - tile->mi_col_starts[tile_col] - 1) &
+ 0x3f;
+ sb_boundary_y_m1 =
+ (tile->mi_row_starts[tile_row + 1] - tile->mi_row_starts[tile_row] - 1) &
+ 0x1ff;
+
+ tile_info_buf[tile_info_base + 3] = (sb_boundary_y_m1 << 7) | sb_boundary_x_m1;
+ tile_info_buf[tile_info_base + 4] = ((allow_update_cdf << 18) | (1 << 16));
+
+ if (tile_num == tile->context_update_tile_id &&
+ uh->disable_frame_end_update_cdf == 0)
+ tile_info_buf[tile_info_base + 4] |= (1 << 17);
+
+ mtk_vcodec_debug(instance, "// tile buf %d pos(%dx%d) offset 0x%x\n",
+ tile_num, tile_row, tile_col, tile_info_base);
+ mtk_vcodec_debug(instance, "// %08x %08x %08x %08x\n",
+ tile_info_buf[tile_info_base + 0],
+ tile_info_buf[tile_info_base + 1],
+ tile_info_buf[tile_info_base + 2],
+ tile_info_buf[tile_info_base + 3]);
+ mtk_vcodec_debug(instance, "// %08x %08x %08x %08x\n",
+ tile_info_buf[tile_info_base + 4],
+ tile_info_buf[tile_info_base + 5],
+ tile_info_buf[tile_info_base + 6],
+ tile_info_buf[tile_info_base + 7]);
+ }
+}
+
+static int vdec_av1_slice_setup_lat(struct vdec_av1_slice_instance *instance,
+ struct mtk_vcodec_mem *bs,
+ struct vdec_lat_buf *lat_buf,
+ struct vdec_av1_slice_pfc *pfc)
+{
+ struct vdec_av1_slice_vsi *vsi = &pfc->vsi;
+ int ret;
+
+ ret = vdec_av1_slice_setup_lat_from_src_buf(instance, vsi, lat_buf);
+ if (ret)
+ return ret;
+
+ ret = vdec_av1_slice_setup_pfc(instance, pfc);
+ if (ret)
+ return ret;
+
+ ret = vdec_av1_slice_setup_tile_group(instance, vsi);
+ if (ret)
+ return ret;
+
+ ret = vdec_av1_slice_alloc_working_buffer(instance, vsi);
+ if (ret)
+ return ret;
+
+ vdec_av1_slice_setup_seg_buffer(instance, vsi);
+ vdec_av1_slice_setup_tile_buffer(instance, vsi, bs);
+ vdec_av1_slice_setup_lat_buffer(instance, vsi, bs, lat_buf);
+
+ return 0;
+}
+
+static int vdec_av1_slice_update_lat(struct vdec_av1_slice_instance *instance,
+ struct vdec_lat_buf *lat_buf,
+ struct vdec_av1_slice_pfc *pfc)
+{
+ struct vdec_av1_slice_vsi *vsi;
+
+ vsi = &pfc->vsi;
+ mtk_vcodec_debug(instance, "frame %u LAT CRC 0x%08x, output size is %d\n",
+ pfc->seq, vsi->state.crc[0], vsi->state.out_size);
+
+ /* buffer full, need to re-decode */
+ if (vsi->state.full) {
+ /* buffer not enough */
+ if (vsi->trans.dma_addr_end - vsi->trans.dma_addr == vsi->ube.size)
+ return -ENOMEM;
+ return -EAGAIN;
+ }
+
+ instance->width = vsi->frame.uh.upscaled_width;
+ instance->height = vsi->frame.uh.frame_height;
+ instance->frame_type = vsi->frame.uh.frame_type;
+
+ return 0;
+}
+
+static int vdec_av1_slice_setup_core_to_dst_buf(struct vdec_av1_slice_instance *instance,
+ struct vdec_lat_buf *lat_buf)
+{
+ struct vb2_v4l2_buffer *dst;
+
+ dst = v4l2_m2m_next_dst_buf(instance->ctx->m2m_ctx);
+ if (!dst)
+ return -EINVAL;
+
+ v4l2_m2m_buf_copy_metadata(&lat_buf->ts_info, dst, true);
+
+ return 0;
+}
+
+static int vdec_av1_slice_setup_core_buffer(struct vdec_av1_slice_instance *instance,
+ struct vdec_av1_slice_pfc *pfc,
+ struct vdec_av1_slice_vsi *vsi,
+ struct vdec_fb *fb,
+ struct vdec_lat_buf *lat_buf)
+{
+ struct vb2_buffer *vb;
+ struct vb2_queue *vq;
+ int w, h, plane, size;
+ int i;
+
+ plane = instance->ctx->q_data[MTK_Q_DATA_DST].fmt->num_planes;
+ w = vsi->frame.uh.upscaled_width;
+ h = vsi->frame.uh.frame_height;
+ size = ALIGN(w, VCODEC_DEC_ALIGNED_64) * ALIGN(h, VCODEC_DEC_ALIGNED_64);
+
+ /* frame buffer */
+ vsi->fb.y.dma_addr = fb->base_y.dma_addr;
+ if (plane == 1)
+ vsi->fb.c.dma_addr = fb->base_y.dma_addr + size;
+ else
+ vsi->fb.c.dma_addr = fb->base_c.dma_addr;
+
+ /* reference buffers */
+ vq = v4l2_m2m_get_vq(instance->ctx->m2m_ctx, V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE);
+ if (!vq)
+ return -EINVAL;
+
+ /* get current output buffer */
+ vb = &v4l2_m2m_next_dst_buf(instance->ctx->m2m_ctx)->vb2_buf;
+ if (!vb)
+ return -EINVAL;
+
+ /* get buffer address from vb2buf */
+ for (i = 0; i < V4L2_AV1_REFS_PER_FRAME; i++) {
+ struct vdec_av1_slice_fb *vref = &vsi->ref[i];
+
+ vb = vb2_find_buffer(vq, pfc->ref_idx[i]);
+ if (!vb) {
+ memset(vref, 0, sizeof(*vref));
+ continue;
+ }
+
+ vref->y.dma_addr = vb2_dma_contig_plane_dma_addr(vb, 0);
+ if (plane == 1)
+ vref->c.dma_addr = vref->y.dma_addr + size;
+ else
+ vref->c.dma_addr = vb2_dma_contig_plane_dma_addr(vb, 1);
+ }
+ vsi->tile.dma_addr = lat_buf->tile_addr.dma_addr;
+ vsi->tile.size = lat_buf->tile_addr.size;
+
+ return 0;
+}
+
+static int vdec_av1_slice_setup_core(struct vdec_av1_slice_instance *instance,
+ struct vdec_fb *fb,
+ struct vdec_lat_buf *lat_buf,
+ struct vdec_av1_slice_pfc *pfc)
+{
+ struct vdec_av1_slice_vsi *vsi = &pfc->vsi;
+ int ret;
+
+ ret = vdec_av1_slice_setup_core_to_dst_buf(instance, lat_buf);
+ if (ret)
+ return ret;
+
+ ret = vdec_av1_slice_setup_core_buffer(instance, pfc, vsi, fb, lat_buf);
+ if (ret)
+ return ret;
+
+ return 0;
+}
+
+static int vdec_av1_slice_update_core(struct vdec_av1_slice_instance *instance,
+ struct vdec_lat_buf *lat_buf,
+ struct vdec_av1_slice_pfc *pfc)
+{
+ struct vdec_av1_slice_vsi *vsi = instance->core_vsi;
+
+ /* TODO: Do something here, or remove this function entirely */
And?
+
+ mtk_vcodec_debug(instance, "frame %u Y_CRC %08x %08x %08x %08x\n",
+ pfc->seq, vsi->state.crc[0], vsi->state.crc[1],
+ vsi->state.crc[2], vsi->state.crc[3]);
+ mtk_vcodec_debug(instance, "frame %u C_CRC %08x %08x %08x %08x\n",
+ pfc->seq, vsi->state.crc[8], vsi->state.crc[9],
+ vsi->state.crc[10], vsi->state.crc[11]);
+
+ return 0;
+}
+
+static int vdec_av1_slice_init(struct mtk_vcodec_ctx *ctx)
+{
+ struct vdec_av1_slice_instance *instance;
+ struct vdec_av1_slice_init_vsi *vsi;
+ int ret;
+
+ instance = kzalloc(sizeof(*instance), GFP_KERNEL);
+ if (!instance)
+ return -ENOMEM;
+
+ instance->ctx = ctx;
+ instance->vpu.id = SCP_IPI_VDEC_LAT;
+ instance->vpu.core_id = SCP_IPI_VDEC_CORE;
+ instance->vpu.ctx = ctx;
+ instance->vpu.codec_type = ctx->current_codec;
+
+ ret = vpu_dec_init(&instance->vpu);
+ if (ret) {
+ mtk_vcodec_err(instance, "failed to init vpu dec, ret %d\n", ret);
+ goto error_vpu_init;
+ }
+
+ /* init vsi and global flags */
+ vsi = instance->vpu.vsi;
+ if (!vsi) {
+ mtk_vcodec_err(instance, "failed to get AV1 vsi\n");
+ ret = -EINVAL;
+ goto error_vsi;
+ }
+ instance->init_vsi = vsi;
+ instance->core_vsi = mtk_vcodec_fw_map_dm_addr(ctx->dev->fw_handler, (u32)vsi->core_vsi);
+
+ if (!instance->core_vsi) {
+ mtk_vcodec_err(instance, "failed to get AV1 core vsi\n");
+ ret = -EINVAL;
+ goto error_vsi;
+ }
+
+ if (vsi->vsi_size != sizeof(struct vdec_av1_slice_vsi))
+ mtk_vcodec_err(instance, "remote vsi size 0x%x mismatch! expected: 0x%lx\n",
+ vsi->vsi_size, sizeof(struct vdec_av1_slice_vsi));
+
+ instance->irq = 1;
Does this mean "irq_enabled"? If so, rename?
+ instance->inneracing_mode = IS_VDEC_INNER_RACING(instance->ctx->dev->dec_capability);
+
+ mtk_vcodec_debug(instance, "vsi 0x%p core_vsi 0x%llx 0x%p, inneracing_mode %d\n",
+ vsi, vsi->core_vsi, instance->core_vsi, instance->inneracing_mode);
+
+ ret = vdec_av1_slice_init_cdf_table(instance);
+ if (ret)
+ goto error_vsi;
+
+ ret = vdec_av1_slice_init_iq_table(instance);
+ if (ret)
+ goto error_vsi;
+
+ ctx->drv_handle = instance;
+
+ return 0;
+error_vsi:
+ vpu_dec_deinit(&instance->vpu);
+error_vpu_init:
+ kfree(instance);
newline?
+ return ret;
+}
+
+static void vdec_av1_slice_deinit(void *h_vdec)
+{
+ struct vdec_av1_slice_instance *instance = h_vdec;
+
+ if (!instance)
+ return;
+ mtk_vcodec_debug(instance, "h_vdec 0x%p\n", h_vdec);
+ vpu_dec_deinit(&instance->vpu);
+ vdec_av1_slice_free_working_buffer(instance);
+ vdec_msg_queue_deinit(&instance->ctx->msg_queue, instance->ctx);
+ kfree(instance);
+}
+
+static int vdec_av1_slice_flush(void *h_vdec, struct mtk_vcodec_mem *bs,
+ struct vdec_fb *fb, bool *res_chg)
+{
+ struct vdec_av1_slice_instance *instance = h_vdec;
+ int i;
+
+ mtk_vcodec_debug(instance, "flush ...\n");
+
+ for (i = 0; i < AV1_MAX_FRAME_BUF_COUNT; i++)
+ vdec_av1_slice_clear_fb(&instance->slots.frame_info[i]);
+
+ vdec_msg_queue_wait_lat_buf_full(&instance->ctx->msg_queue);
newline?
+ return vpu_dec_reset(&instance->vpu);
+}
+
+static void vdec_av1_slice_get_pic_info(struct vdec_av1_slice_instance *instance)
+{
+ struct mtk_vcodec_ctx *ctx = instance->ctx;
+ u32 data[3];
+
+ mtk_vcodec_debug(instance, "w %u h %u\n", ctx->picinfo.pic_w, ctx->picinfo.pic_h);
+
+ data[0] = ctx->picinfo.pic_w;
+ data[1] = ctx->picinfo.pic_h;
+ data[2] = ctx->capture_fourcc;
+ vpu_dec_get_param(&instance->vpu, data, 3, GET_PARAM_PIC_INFO);
+
+ ctx->picinfo.buf_w = ALIGN(ctx->picinfo.pic_w, VCODEC_DEC_ALIGNED_64);
+ ctx->picinfo.buf_h = ALIGN(ctx->picinfo.pic_h, VCODEC_DEC_ALIGNED_64);
+ ctx->picinfo.fb_sz[0] = instance->vpu.fb_sz[0];
+ ctx->picinfo.fb_sz[1] = instance->vpu.fb_sz[1];
+}
+
+static void vdec_av1_slice_get_dpb_size(struct vdec_av1_slice_instance *instance, u32 *dpb_sz)
static inline void?
+{
+ /* refer av1 specification */
+ *dpb_sz = V4L2_AV1_TOTAL_REFS_PER_FRAME + 1;
+}
+
+static void vdec_av1_slice_get_crop_info(struct vdec_av1_slice_instance *instance,
+ struct v4l2_rect *cr)
+{
+ struct mtk_vcodec_ctx *ctx = instance->ctx;
+
+ cr->left = 0;
+ cr->top = 0;
+ cr->width = ctx->picinfo.pic_w;
+ cr->height = ctx->picinfo.pic_h;
+
+ mtk_vcodec_debug(instance, "l=%d, t=%d, w=%d, h=%d\n",
+ cr->left, cr->top, cr->width, cr->height);
+}
+
+static int vdec_av1_slice_get_param(void *h_vdec, enum vdec_get_param_type type, void *out)
+{
+ struct vdec_av1_slice_instance *instance = h_vdec;
+
+ switch (type) {
+ case GET_PARAM_PIC_INFO:
+ vdec_av1_slice_get_pic_info(instance);
+ break;
+ case GET_PARAM_DPB_SIZE:
+ vdec_av1_slice_get_dpb_size(instance, out);
+ break;
+ case GET_PARAM_CROP_INFO:
+ vdec_av1_slice_get_crop_info(instance, out);
+ break;
+ default:
+ mtk_vcodec_err(instance, "invalid get parameter type=%d\n", type);
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+static int vdec_av1_slice_lat_decode(void *h_vdec, struct mtk_vcodec_mem *bs,
+ struct vdec_fb *fb, bool *res_chg)
+{
+ struct vdec_av1_slice_instance *instance = h_vdec;
+ struct vdec_lat_buf *lat_buf;
+ struct vdec_av1_slice_pfc *pfc;
+ struct vdec_av1_slice_vsi *vsi;
+ struct mtk_vcodec_ctx *ctx;
+ int ret;
+
+ if (!instance || !instance->ctx)
+ return -EINVAL;
+
+ ctx = instance->ctx;
+ /* init msgQ for the first time */
+ if (vdec_msg_queue_init(&ctx->msg_queue, ctx,
+ vdec_av1_slice_core_decode, sizeof(*pfc))) {
+ mtk_vcodec_err(instance, "failed to init AV1 msg queue\n");
+ return -ENOMEM;
+ }
+
+ /* bs NULL means flush decoder */
+ if (!bs)
+ return vdec_av1_slice_flush(h_vdec, bs, fb, res_chg);
+
+ lat_buf = vdec_msg_queue_dqbuf(&ctx->msg_queue.lat_ctx);
+ if (!lat_buf) {
+ mtk_vcodec_err(instance, "failed to get AV1 lat buf\n");
there exists vdec_msg_queue_deinit(). Should it be called in this (and
subsequent) error recovery path(s)?
+ return -EBUSY;
+ }
+ pfc = (struct vdec_av1_slice_pfc *)lat_buf->private_data;
+ if (!pfc) {
+ ret = -EINVAL;
+ goto err_free_fb_out;
+ }
+ vsi = &pfc->vsi;
+
+ ret = vdec_av1_slice_setup_lat(instance, bs, lat_buf, pfc);
+ if (ret) {
+ mtk_vcodec_err(instance, "failed to setup AV1 lat ret %d\n", ret);
+ goto err_free_fb_out;
+ }
+
+ vdec_av1_slice_vsi_to_remote(vsi, instance->vsi);
+ ret = vpu_dec_start(&instance->vpu, NULL, 0);
+ if (ret) {
+ mtk_vcodec_err(instance, "failed to dec AV1 ret %d\n", ret);
+ goto err_free_fb_out;
+ }
+ if (instance->inneracing_mode)
+ vdec_msg_queue_qbuf(&ctx->dev->msg_queue_core_ctx, lat_buf);
+
+ if (instance->irq) {
+ ret = mtk_vcodec_wait_for_done_ctx(ctx, MTK_INST_IRQ_RECEIVED,
+ WAIT_INTR_TIMEOUT_MS,
+ MTK_VDEC_LAT0);
+ /* update remote vsi if decode timeout */
+ if (ret) {
+ mtk_vcodec_err(instance, "AV1 Frame %d decode timeout %d\n", pfc->seq, ret);
+ WRITE_ONCE(instance->vsi->state.timeout, 1);
+ }
+ vpu_dec_end(&instance->vpu);
+ }
+
+ vdec_av1_slice_vsi_from_remote(vsi, instance->vsi);
+ ret = vdec_av1_slice_update_lat(instance, lat_buf, pfc);
+
+ /* LAT trans full, re-decode */
+ if (ret == -EAGAIN) {
+ mtk_vcodec_err(instance, "AV1 Frame %d trans full\n", pfc->seq);
+ if (!instance->inneracing_mode)
+ vdec_msg_queue_qbuf(&ctx->msg_queue.lat_ctx, lat_buf);
+ return 0;
+ }
+
+ /* LAT trans full, no more UBE or decode timeout */
+ if (ret == -ENOMEM || vsi->state.timeout) {
+ mtk_vcodec_err(instance, "AV1 Frame %d insufficient buffer or timeout\n", pfc->seq);
+ if (!instance->inneracing_mode)
+ vdec_msg_queue_qbuf(&ctx->msg_queue.lat_ctx, lat_buf);
+ return -EBUSY;
+ }
+ vsi->trans.dma_addr_end += ctx->msg_queue.wdma_addr.dma_addr;
+ mtk_vcodec_debug(instance, "lat dma 1 0x%llx 0x%llx\n",
+ pfc->vsi.trans.dma_addr, pfc->vsi.trans.dma_addr_end);
+
+ vdec_msg_queue_update_ube_wptr(&ctx->msg_queue, vsi->trans.dma_addr_end);
+
+ if (!instance->inneracing_mode)
+ vdec_msg_queue_qbuf(&ctx->dev->msg_queue_core_ctx, lat_buf);
+ memcpy(&instance->slots, &vsi->slots, sizeof(instance->slots));
+
+ return 0;
+
+err_free_fb_out:
+ vdec_msg_queue_qbuf(&ctx->msg_queue.lat_ctx, lat_buf);
+ mtk_vcodec_err(instance, "slice dec number: %d err: %d", pfc->seq, ret);
+ return ret;
+}
+
+static int vdec_av1_slice_core_decode(struct vdec_lat_buf *lat_buf)
+{
+ struct vdec_av1_slice_instance *instance;
+ struct vdec_av1_slice_pfc *pfc;
+ struct mtk_vcodec_ctx *ctx = NULL;
+ struct vdec_fb *fb = NULL;
+ int ret = -EINVAL;
+
+ if (!lat_buf)
+ return -EINVAL;
+
+ pfc = lat_buf->private_data;
+ ctx = lat_buf->ctx;
+ if (!pfc || !ctx)
+ return -EINVAL;
+
+ instance = ctx->drv_handle;
+ if (!instance)
+ goto err;
+
+ fb = ctx->dev->vdec_pdata->get_cap_buffer(ctx);
+ if (!fb) {
+ ret = -EBUSY;
+ goto err;
+ }
+
+ ret = vdec_av1_slice_setup_core(instance, fb, lat_buf, pfc);
+ if (ret) {
+ mtk_vcodec_err(instance, "vdec_av1_slice_setup_core\n");
+ goto err;
+ }
+ vdec_av1_slice_vsi_to_remote(&pfc->vsi, instance->core_vsi);
+ ret = vpu_dec_core(&instance->vpu);
+ if (ret) {
+ mtk_vcodec_err(instance, "vpu_dec_core\n");
+ goto err;
+ }
+
+ if (instance->irq) {
+ ret = mtk_vcodec_wait_for_done_ctx(ctx, MTK_INST_IRQ_RECEIVED,
+ WAIT_INTR_TIMEOUT_MS,
+ MTK_VDEC_CORE);
+ /* update remote vsi if decode timeout */
+ if (ret) {
+ mtk_vcodec_err(instance, "AV1 frame %d core timeout\n", pfc->seq);
+ WRITE_ONCE(instance->vsi->state.timeout, 1);
+ }
+ vpu_dec_core_end(&instance->vpu);
+ }
+
+ ret = vdec_av1_slice_update_core(instance, lat_buf, pfc);
+ if (ret) {
+ mtk_vcodec_err(instance, "vdec_av1_slice_update_core\n");
+ goto err;
+ }
+
+ mtk_vcodec_debug(instance, "core dma_addr_end 0x%llx\n",
+ instance->core_vsi->trans.dma_addr_end);
+ vdec_msg_queue_update_ube_rptr(&ctx->msg_queue, instance->core_vsi->trans.dma_addr_end);
+
+ ctx->dev->vdec_pdata->cap_to_disp(ctx, 0, lat_buf->src_buf_req);
+
+ return 0;
+
+err:
+ /* always update read pointer */
+ vdec_msg_queue_update_ube_rptr(&ctx->msg_queue, pfc->vsi.trans.dma_addr_end);
+
+ if (fb)
+ ctx->dev->vdec_pdata->cap_to_disp(ctx, 1, lat_buf->src_buf_req);
+
+ return ret;
+}
+
+const struct vdec_common_if vdec_av1_slice_lat_if = {
+ .init = vdec_av1_slice_init,
+ .decode = vdec_av1_slice_lat_decode,
+ .get_param = vdec_av1_slice_get_param,
+ .deinit = vdec_av1_slice_deinit,
+};
diff --git a/drivers/media/platform/mediatek/vcodec/vdec_drv_if.c b/drivers/media/platform/mediatek/vcodec/vdec_drv_if.c
index f3807f03d8806..4dda59a6c8141 100644
--- a/drivers/media/platform/mediatek/vcodec/vdec_drv_if.c
+++ b/drivers/media/platform/mediatek/vcodec/vdec_drv_if.c
@@ -49,6 +49,10 @@ int vdec_if_init(struct mtk_vcodec_ctx *ctx, unsigned int fourcc)
ctx->dec_if = &vdec_vp9_slice_lat_if;
ctx->hw_id = IS_VDEC_LAT_ARCH(hw_arch) ? MTK_VDEC_LAT0 : MTK_VDEC_CORE;
break;
+ case V4L2_PIX_FMT_AV1_FRAME:
+ ctx->dec_if = &vdec_av1_slice_lat_if;
+ ctx->hw_id = MTK_VDEC_LAT0;
+ break;
default:
return -EINVAL;
}
diff --git a/drivers/media/platform/mediatek/vcodec/vdec_drv_if.h b/drivers/media/platform/mediatek/vcodec/vdec_drv_if.h
index 076306ff2dd49..dc6c8ecd9843a 100644
--- a/drivers/media/platform/mediatek/vcodec/vdec_drv_if.h
+++ b/drivers/media/platform/mediatek/vcodec/vdec_drv_if.h
@@ -61,6 +61,7 @@ extern const struct vdec_common_if vdec_vp8_if;
extern const struct vdec_common_if vdec_vp8_slice_if;
extern const struct vdec_common_if vdec_vp9_if;
extern const struct vdec_common_if vdec_vp9_slice_lat_if;
+extern const struct vdec_common_if vdec_av1_slice_lat_if;
/**
* vdec_if_init() - initialize decode driver
diff --git a/drivers/media/platform/mediatek/vcodec/vdec_msg_queue.c b/drivers/media/platform/mediatek/vcodec/vdec_msg_queue.c
index ae500980ad45c..05b54b0e3f2d2 100644
--- a/drivers/media/platform/mediatek/vcodec/vdec_msg_queue.c
+++ b/drivers/media/platform/mediatek/vcodec/vdec_msg_queue.c
@@ -20,6 +20,9 @@
/* the size used to store avc error information */
#define VDEC_ERR_MAP_SZ_AVC (17 * SZ_1K)
+#define VDEC_RD_MV_BUFFER_SZ (((SZ_4K * 2304 >> 4) + SZ_1K) << 1)
+#define VDEC_LAT_TILE_SZ (64 * SZ_4K)
+
/* core will read the trans buffer which decoded by lat to decode again.
* The trans buffer size of FHD and 4K bitstreams are different.
*/
@@ -194,6 +197,14 @@ void vdec_msg_queue_deinit(struct vdec_msg_queue *msg_queue,
if (mem->va)
mtk_vcodec_mem_free(ctx, mem);
+ mem = &lat_buf->rd_mv_addr;
+ if (mem->va)
+ mtk_vcodec_mem_free(ctx, mem);
+
+ mem = &lat_buf->tile_addr;
+ if (mem->va)
+ mtk_vcodec_mem_free(ctx, mem);
+
kfree(lat_buf->private_data);
}
}
@@ -270,6 +281,22 @@ int vdec_msg_queue_init(struct vdec_msg_queue *msg_queue,
goto mem_alloc_err;
}
+ if (ctx->current_codec == V4L2_PIX_FMT_AV1_FRAME) {
+ lat_buf->rd_mv_addr.size = VDEC_RD_MV_BUFFER_SZ;
+ err = mtk_vcodec_mem_alloc(ctx, &lat_buf->rd_mv_addr);
+ if (err) {
+ mtk_v4l2_err("failed to allocate rd_mv_addr buf[%d]", i);
+ return -ENOMEM;
+ }
+
+ lat_buf->tile_addr.size = VDEC_LAT_TILE_SZ;
+ err = mtk_vcodec_mem_alloc(ctx, &lat_buf->tile_addr);
+ if (err) {
+ mtk_v4l2_err("failed to allocate tile_addr buf[%d]", i);
+ return -ENOMEM;
+ }
+ }
+
lat_buf->private_data = kzalloc(private_size, GFP_KERNEL);
if (!lat_buf->private_data) {
err = -ENOMEM;
diff --git a/drivers/media/platform/mediatek/vcodec/vdec_msg_queue.h b/drivers/media/platform/mediatek/vcodec/vdec_msg_queue.h
index c43d427f5f544..525170e411ee0 100644
--- a/drivers/media/platform/mediatek/vcodec/vdec_msg_queue.h
+++ b/drivers/media/platform/mediatek/vcodec/vdec_msg_queue.h
@@ -42,6 +42,8 @@ struct vdec_msg_queue_ctx {
* struct vdec_lat_buf - lat buffer message used to store lat info for core decode
* @wdma_err_addr: wdma error address used for lat hardware
* @slice_bc_addr: slice bc address used for lat hardware
+ * @rd_mv_addr: mv addr for av1 lat hardware output, core hardware input
+ * @tile_addr: tile buffer for av1 core input
* @ts_info: need to set timestamp from output to capture
* @src_buf_req: output buffer media request object
*
@@ -54,6 +56,8 @@ struct vdec_msg_queue_ctx {
struct vdec_lat_buf {
struct mtk_vcodec_mem wdma_err_addr;
struct mtk_vcodec_mem slice_bc_addr;
+ struct mtk_vcodec_mem rd_mv_addr;
+ struct mtk_vcodec_mem tile_addr;
struct vb2_v4l2_buffer ts_info;
struct media_request *src_buf_req;