Linux-Rockchip Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Benjamin Gaignard <benjamin.gaignard@collabora.com>
To: nicolas.dufresne@collabora.com, p.zabel@pengutronix.de,
	mchehab@kernel.org, heiko@sntech.de
Cc: linux-kernel@vger.kernel.org, linux-media@vger.kernel.org,
	linux-rockchip@lists.infradead.org,
	linux-arm-kernel@lists.infradead.org, kernel@collabora.com
Subject: Re: [PATCH v5] media: verisilicon: Create AV1 helper library
Date: Mon, 31 Aug 2026 15:38:03 +0200	[thread overview]
Message-ID: <25e17f59-e637-48a2-8a9b-4913738b7c4e@collabora.com> (raw)
In-Reply-To: <6eb8768e-e46c-47f4-966d-bfc05c8a6fe4@collabora.com>


Le 10/07/2026 à 11:45, Benjamin Gaignard a écrit :
> Gentle ping on this one.
>
> Regards,
> Benjamin
>
> Le 12/05/2026 à 09:59, Benjamin Gaignard a écrit :
>> Regroup all none hardware related AV1 functions into a helper library.
>> The goal is to avoid code duplication for future AV1 codecs.
>>
>> Tested on rock 5b board Fluster score remains the same 204/241.
>>
>> Signed-off-by: Benjamin Gaignard <benjamin.gaignard@collabora.com>
>> Reviewed-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>

Hi Hans,

Maybe something is missing or blocking this patch but I don't know what.
I would like to progress on it because I plan to do other improvements
for Verisilicon drivers and they depend of this one.
If there any I can do to unlock this situation please tell me.

Thanks for your time.
Regards,
Benjamin

>> ---
>> version 5:
>> - be more strict about checking arrays range and the value returned
>>    by hantro_av1_get_frame_index()
>> - rebased on v7.1-rc3 tag
>>
>> version 4:
>> - change functions prototypes to use a structure.
>> - rebased on v7.1-rc2 tag.
>>
>>   drivers/media/platform/verisilicon/Makefile   |    7 +-
>>   .../media/platform/verisilicon/hantro_av1.c   |  790 +++++++++++++
>>   .../media/platform/verisilicon/hantro_av1.h   |   64 ++
>>   ...entropymode.c => hantro_av1_entropymode.c} |   18 +-
>>   ...entropymode.h => hantro_av1_entropymode.h} |   18 +-
>>   ...av1_filmgrain.c => hantro_av1_filmgrain.c} |  125 +-
>>   .../verisilicon/hantro_av1_filmgrain.h        |   48 +
>>   .../media/platform/verisilicon/hantro_hw.h    |    7 +-
>>   .../verisilicon/rockchip_av1_filmgrain.h      |   36 -
>>   .../verisilicon/rockchip_vpu981_hw_av1_dec.c  | 1014 ++---------------
>>   .../platform/verisilicon/rockchip_vpu_hw.c    |    7 +-
>>   11 files changed, 1115 insertions(+), 1019 deletions(-)
>>   create mode 100644 drivers/media/platform/verisilicon/hantro_av1.c
>>   create mode 100644 drivers/media/platform/verisilicon/hantro_av1.h
>>   rename 
>> drivers/media/platform/verisilicon/{rockchip_av1_entropymode.c => 
>> hantro_av1_entropymode.c} (99%)
>>   rename 
>> drivers/media/platform/verisilicon/{rockchip_av1_entropymode.h => 
>> hantro_av1_entropymode.h} (95%)
>>   rename drivers/media/platform/verisilicon/{rockchip_av1_filmgrain.c 
>> => hantro_av1_filmgrain.c} (85%)
>>   create mode 100644 
>> drivers/media/platform/verisilicon/hantro_av1_filmgrain.h
>>   delete mode 100644 
>> drivers/media/platform/verisilicon/rockchip_av1_filmgrain.h
>>
>> diff --git a/drivers/media/platform/verisilicon/Makefile 
>> b/drivers/media/platform/verisilicon/Makefile
>> index f6f019d04ff0..a1dd6c2d29be 100644
>> --- a/drivers/media/platform/verisilicon/Makefile
>> +++ b/drivers/media/platform/verisilicon/Makefile
>> @@ -19,7 +19,10 @@ hantro-vpu-y += \
>>           hantro_hevc.o \
>>           hantro_mpeg2.o \
>>           hantro_vp8.o \
>> -        hantro_vp9.o
>> +        hantro_vp9.o \
>> +        hantro_av1.o \
>> +        hantro_av1_filmgrain.o \
>> +        hantro_av1_entropymode.o
>>     hantro-vpu-$(CONFIG_VIDEO_HANTRO_IMX8M) += \
>>           imx8m_vpu_hw.o
>> @@ -33,8 +36,6 @@ hantro-vpu-$(CONFIG_VIDEO_HANTRO_ROCKCHIP) += \
>>           rockchip_vpu2_hw_mpeg2_dec.o \
>>           rockchip_vpu2_hw_vp8_dec.o \
>>           rockchip_vpu981_hw_av1_dec.o \
>> -        rockchip_av1_filmgrain.o \
>> -        rockchip_av1_entropymode.o \
>>           rockchip_vpu_hw.o
>>     hantro-vpu-$(CONFIG_VIDEO_HANTRO_SUNXI) += \
>> diff --git a/drivers/media/platform/verisilicon/hantro_av1.c 
>> b/drivers/media/platform/verisilicon/hantro_av1.c
>> new file mode 100644
>> index 000000000000..ac7ae8e87f50
>> --- /dev/null
>> +++ b/drivers/media/platform/verisilicon/hantro_av1.c
>> @@ -0,0 +1,790 @@
>> +// SPDX-License-Identifier: GPL-2.0-only
>> +/*
>> + * Copyright (c) 2026, Collabora
>> + *
>> + * Author: Benjamin Gaignard <benjamin.gaignard@collabora.com>
>> + */
>> +
>> +#include <linux/types.h>
>> +#include <media/v4l2-h264.h>
>> +#include <media/v4l2-mem2mem.h>
>> +
>> +#include "hantro.h"
>> +#include "hantro_av1.h"
>> +#include "hantro_hw.h"
>> +
>> +#define GM_GLOBAL_MODELS_PER_FRAME    7
>> +#define GLOBAL_MODEL_TOTAL_SIZE    (6 * 4 + 4 * 2)
>> +#define GLOBAL_MODEL_SIZE    ALIGN(GM_GLOBAL_MODELS_PER_FRAME * 
>> GLOBAL_MODEL_TOTAL_SIZE, 2048)
>> +#define AV1_MAX_TILES        128
>> +#define AV1_TILE_INFO_SIZE    (AV1_MAX_TILES * 16)
>> +#define AV1_TILE_SIZE        ALIGN(32 * 128, 4096)
>> +
>> +#define SUPERRES_SCALE_BITS 3
>> +
>> +#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 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_)));        \
>> +})
>> +
>> +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,
>> +};
>> +
>> +enum hantro_av1_tx_mode {
>> +    HANTRO_AV1_TX_MODE_ONLY_4X4    = 0,
>> +    HANTRO_AV1_TX_MODE_8X8        = 1,
>> +    HANTRO_AV1_TX_MODE_16x16    = 2,
>> +    HANTRO_AV1_TX_MODE_32x32    = 3,
>> +    HANTRO_AV1_TX_MODE_SELECT    = 4,
>> +};
>> +
>> +enum hantro_av1_inter_prediction_filter_type {
>> +    HANTRO_AV1_EIGHT_TAP_SMOOTH    = 0,
>> +    HANTRO_AV1_EIGHT_TAP        = 1,
>> +    HANTRO_AV1_EIGHT_TAP_SHARP    = 2,
>> +    HANTRO_AV1_BILINEAR        = 3,
>> +    HANTRO_AV1_SWITCHABLE        = 4,
>> +};
>> +
>> +int hantro_av1_get_hardware_tx_mode(enum v4l2_av1_tx_mode tx_mode)
>> +{
>> +    switch (tx_mode) {
>> +    case V4L2_AV1_TX_MODE_ONLY_4X4:
>> +        return HANTRO_AV1_TX_MODE_ONLY_4X4;
>> +    case V4L2_AV1_TX_MODE_LARGEST:
>> +        return HANTRO_AV1_TX_MODE_32x32;
>> +    case V4L2_AV1_TX_MODE_SELECT:
>> +        return HANTRO_AV1_TX_MODE_SELECT;
>> +    }
>> +
>> +    return HANTRO_AV1_TX_MODE_32x32;
>> +}
>> +
>> +int hantro_av1_get_hardware_mcomp_filt_type(int interpolation_filter)
>> +{
>> +    switch (interpolation_filter) {
>> +    case V4L2_AV1_INTERPOLATION_FILTER_EIGHTTAP:
>> +        return HANTRO_AV1_EIGHT_TAP;
>> +    case V4L2_AV1_INTERPOLATION_FILTER_EIGHTTAP_SMOOTH:
>> +        return HANTRO_AV1_EIGHT_TAP_SMOOTH;
>> +    case V4L2_AV1_INTERPOLATION_FILTER_EIGHTTAP_SHARP:
>> +        return HANTRO_AV1_EIGHT_TAP_SHARP;
>> +    case V4L2_AV1_INTERPOLATION_FILTER_BILINEAR:
>> +        return HANTRO_AV1_BILINEAR;
>> +    case V4L2_AV1_INTERPOLATION_FILTER_SWITCHABLE:
>> +        return HANTRO_AV1_SWITCHABLE;
>> +    }
>> +
>> +    return HANTRO_AV1_EIGHT_TAP_SMOOTH;
>> +}
>> +
>> +int hantro_av1_get_frame_index(struct hantro_ctx *ctx, int ref)
>> +{
>> +    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> +    struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>> +    const struct v4l2_ctrl_av1_frame *frame = ctrls->frame;
>> +    u64 timestamp;
>> +    int i, idx;
>> +
>> +    if (ref < 0 || ref >= V4L2_AV1_REFS_PER_FRAME)
>> +        return AV1_INVALID_IDX;
>> +
>> +    idx = frame->ref_frame_idx[ref];
>> +
>> +    if (idx >= V4L2_AV1_TOTAL_REFS_PER_FRAME || idx < 0)
>> +        return AV1_INVALID_IDX;
>> +
>> +    timestamp = frame->reference_frame_ts[idx];
>> +    for (i = 0; i < AV1_MAX_FRAME_BUF_COUNT; i++) {
>> +        if (!av1_dec->frame_refs[i].used)
>> +            continue;
>> +        if (av1_dec->frame_refs[i].timestamp == timestamp)
>> +            return i;
>> +    }
>> +
>> +    return AV1_INVALID_IDX;
>> +}
>> +
>> +int hantro_av1_get_order_hint(struct hantro_ctx *ctx, int ref)
>> +{
>> +    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> +    int idx = hantro_av1_get_frame_index(ctx, ref);
>> +
>> +    if (idx != AV1_INVALID_IDX)
>> +        return av1_dec->frame_refs[idx].order_hint;
>> +
>> +    return 0;
>> +}
>> +
>> +int hantro_av1_frame_ref(struct hantro_ctx *ctx, u64 timestamp)
>> +{
>> +    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> +    struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>> +    const struct v4l2_ctrl_av1_frame *frame = ctrls->frame;
>> +    int i;
>> +
>> +    for (i = 0; i < AV1_MAX_FRAME_BUF_COUNT; i++) {
>> +        int j;
>> +
>> +        if (av1_dec->frame_refs[i].used)
>> +            continue;
>> +
>> +        av1_dec->frame_refs[i].width = frame->frame_width_minus_1 + 1;
>> +        av1_dec->frame_refs[i].height = frame->frame_height_minus_1 
>> + 1;
>> +        av1_dec->frame_refs[i].mi_cols = 
>> DIV_ROUND_UP(frame->frame_width_minus_1 + 1, 8);
>> +        av1_dec->frame_refs[i].mi_rows = 
>> DIV_ROUND_UP(frame->frame_height_minus_1 + 1, 8);
>> +        av1_dec->frame_refs[i].timestamp = timestamp;
>> +        av1_dec->frame_refs[i].frame_type = frame->frame_type;
>> +        av1_dec->frame_refs[i].order_hint = frame->order_hint;
>> +        av1_dec->frame_refs[i].vb2_ref = hantro_get_dst_buf(ctx);
>> +
>> +        for (j = 0; j < V4L2_AV1_TOTAL_REFS_PER_FRAME; j++)
>> +            av1_dec->frame_refs[i].order_hints[j] = 
>> frame->order_hints[j];
>> +        av1_dec->frame_refs[i].used = true;
>> +        av1_dec->current_frame_index = i;
>> +
>> +        return i;
>> +    }
>> +
>> +    return AV1_INVALID_IDX;
>> +}
>> +
>> +static void hantro_av1_frame_unref(struct hantro_ctx *ctx, int idx)
>> +{
>> +    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> +
>> +    if (idx >= 0)
>> +        av1_dec->frame_refs[idx].used = false;
>> +}
>> +
>> +void hantro_av1_clean_refs(struct hantro_ctx *ctx)
>> +{
>> +    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> +    struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>> +
>> +    int ref, idx;
>> +
>> +    for (idx = 0; idx < AV1_MAX_FRAME_BUF_COUNT; idx++) {
>> +        u64 timestamp = av1_dec->frame_refs[idx].timestamp;
>> +        bool used = false;
>> +
>> +        if (!av1_dec->frame_refs[idx].used)
>> +            continue;
>> +
>> +        for (ref = 0; ref < V4L2_AV1_TOTAL_REFS_PER_FRAME; ref++) {
>> +            if (ctrls->frame->reference_frame_ts[ref] == timestamp)
>> +                used = true;
>> +        }
>> +
>> +        if (!used)
>> +            hantro_av1_frame_unref(ctx, idx);
>> +    }
>> +}
>> +
>> +size_t hantro_av1_luma_size(struct hantro_ctx *ctx)
>> +{
>> +    return ctx->ref_fmt.plane_fmt[0].bytesperline * 
>> ctx->ref_fmt.height;
>> +}
>> +
>> +size_t hantro_av1_chroma_size(struct hantro_ctx *ctx)
>> +{
>> +    size_t cr_offset = hantro_av1_luma_size(ctx);
>> +
>> +    return ALIGN((cr_offset * 3) / 2, 64);
>> +}
>> +
>> +static void hantro_av1_tiles_free(struct hantro_ctx *ctx)
>> +{
>> +    struct hantro_dev *vpu = ctx->dev;
>> +    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> +
>> +    if (av1_dec->db_data_col.cpu)
>> +        dma_free_coherent(vpu->dev, av1_dec->db_data_col.size,
>> +                  av1_dec->db_data_col.cpu,
>> +                  av1_dec->db_data_col.dma);
>> +    av1_dec->db_data_col.cpu = NULL;
>> +
>> +    if (av1_dec->db_ctrl_col.cpu)
>> +        dma_free_coherent(vpu->dev, av1_dec->db_ctrl_col.size,
>> +                  av1_dec->db_ctrl_col.cpu,
>> +                  av1_dec->db_ctrl_col.dma);
>> +    av1_dec->db_ctrl_col.cpu = NULL;
>> +
>> +    if (av1_dec->cdef_col.cpu)
>> +        dma_free_coherent(vpu->dev, av1_dec->cdef_col.size,
>> +                  av1_dec->cdef_col.cpu, av1_dec->cdef_col.dma);
>> +    av1_dec->cdef_col.cpu = NULL;
>> +
>> +    if (av1_dec->sr_col.cpu)
>> +        dma_free_coherent(vpu->dev, av1_dec->sr_col.size,
>> +                  av1_dec->sr_col.cpu, av1_dec->sr_col.dma);
>> +    av1_dec->sr_col.cpu = NULL;
>> +
>> +    if (av1_dec->lr_col.cpu)
>> +        dma_free_coherent(vpu->dev, av1_dec->lr_col.size,
>> +                  av1_dec->lr_col.cpu, av1_dec->lr_col.dma);
>> +    av1_dec->lr_col.cpu = NULL;
>> +}
>> +
>> +static int hantro_av1_tiles_reallocate(struct hantro_ctx *ctx)
>> +{
>> +    struct hantro_dev *vpu = ctx->dev;
>> +    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> +    struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>> +    const struct v4l2_av1_tile_info *tile_info = 
>> &ctrls->frame->tile_info;
>> +    unsigned int num_tile_cols = tile_info->tile_cols;
>> +    unsigned int height = ALIGN(ctrls->frame->frame_height_minus_1 + 
>> 1, 64);
>> +    unsigned int height_in_sb = height / 64;
>> +    unsigned int stripe_num = ((height + 8) + 63) / 64;
>> +    size_t size;
>> +
>> +    if (av1_dec->db_data_col.size >=
>> +        ALIGN(height * 12 * ctx->bit_depth / 8, 128) * num_tile_cols)
>> +        return 0;
>> +
>> +    hantro_av1_tiles_free(ctx);
>> +
>> +    size = ALIGN(height * 12 * ctx->bit_depth / 8, 128) * 
>> num_tile_cols;
>> +    av1_dec->db_data_col.cpu = dma_alloc_coherent(vpu->dev, size,
>> +                              &av1_dec->db_data_col.dma,
>> +                              GFP_KERNEL);
>> +    if (!av1_dec->db_data_col.cpu)
>> +        goto buffer_allocation_error;
>> +    av1_dec->db_data_col.size = size;
>> +
>> +    size = ALIGN(height * 2 * 16 / 4, 128) * num_tile_cols;
>> +    av1_dec->db_ctrl_col.cpu = dma_alloc_coherent(vpu->dev, size,
>> +                              &av1_dec->db_ctrl_col.dma,
>> +                              GFP_KERNEL);
>> +    if (!av1_dec->db_ctrl_col.cpu)
>> +        goto buffer_allocation_error;
>> +    av1_dec->db_ctrl_col.size = size;
>> +
>> +    size = ALIGN(height_in_sb * 44 * ctx->bit_depth * 16 / 8, 128) * 
>> num_tile_cols;
>> +    av1_dec->cdef_col.cpu = dma_alloc_coherent(vpu->dev, size,
>> +                           &av1_dec->cdef_col.dma,
>> +                           GFP_KERNEL);
>> +    if (!av1_dec->cdef_col.cpu)
>> +        goto buffer_allocation_error;
>> +    av1_dec->cdef_col.size = size;
>> +
>> +    size = ALIGN(height_in_sb * (3040 + 1280), 128) * num_tile_cols;
>> +    av1_dec->sr_col.cpu = dma_alloc_coherent(vpu->dev, size,
>> +                         &av1_dec->sr_col.dma,
>> +                         GFP_KERNEL);
>> +    if (!av1_dec->sr_col.cpu)
>> +        goto buffer_allocation_error;
>> +    av1_dec->sr_col.size = size;
>> +
>> +    size = ALIGN(stripe_num * 1536 * ctx->bit_depth / 8, 128) * 
>> num_tile_cols;
>> +    av1_dec->lr_col.cpu = dma_alloc_coherent(vpu->dev, size,
>> +                         &av1_dec->lr_col.dma,
>> +                         GFP_KERNEL);
>> +    if (!av1_dec->lr_col.cpu)
>> +        goto buffer_allocation_error;
>> +    av1_dec->lr_col.size = size;
>> +
>> +    av1_dec->num_tile_cols_allocated = num_tile_cols;
>> +    return 0;
>> +
>> +buffer_allocation_error:
>> +    hantro_av1_tiles_free(ctx);
>> +    return -ENOMEM;
>> +}
>> +
>> +void hantro_av1_exit(struct hantro_ctx *ctx)
>> +{
>> +    struct hantro_dev *vpu = ctx->dev;
>> +    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> +
>> +    if (av1_dec->global_model.cpu)
>> +        dma_free_coherent(vpu->dev, av1_dec->global_model.size,
>> +                  av1_dec->global_model.cpu,
>> +                  av1_dec->global_model.dma);
>> +    av1_dec->global_model.cpu = NULL;
>> +
>> +    if (av1_dec->tile_info.cpu)
>> +        dma_free_coherent(vpu->dev, av1_dec->tile_info.size,
>> +                  av1_dec->tile_info.cpu,
>> +                  av1_dec->tile_info.dma);
>> +    av1_dec->tile_info.cpu = NULL;
>> +
>> +    if (av1_dec->film_grain.cpu)
>> +        dma_free_coherent(vpu->dev, av1_dec->film_grain.size,
>> +                  av1_dec->film_grain.cpu,
>> +                  av1_dec->film_grain.dma);
>> +    av1_dec->film_grain.cpu = NULL;
>> +
>> +    if (av1_dec->prob_tbl.cpu)
>> +        dma_free_coherent(vpu->dev, av1_dec->prob_tbl.size,
>> +                  av1_dec->prob_tbl.cpu, av1_dec->prob_tbl.dma);
>> +    av1_dec->prob_tbl.cpu = NULL;
>> +
>> +    if (av1_dec->prob_tbl_out.cpu)
>> +        dma_free_coherent(vpu->dev, av1_dec->prob_tbl_out.size,
>> +                  av1_dec->prob_tbl_out.cpu,
>> +                  av1_dec->prob_tbl_out.dma);
>> +    av1_dec->prob_tbl_out.cpu = NULL;
>> +
>> +    if (av1_dec->tile_buf.cpu)
>> +        dma_free_coherent(vpu->dev, av1_dec->tile_buf.size,
>> +                  av1_dec->tile_buf.cpu, av1_dec->tile_buf.dma);
>> +    av1_dec->tile_buf.cpu = NULL;
>> +
>> +    hantro_av1_tiles_free(ctx);
>> +}
>> +
>> +int hantro_av1_init(struct hantro_ctx *ctx)
>> +{
>> +    struct hantro_dev *vpu = ctx->dev;
>> +    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> +
>> +    memset(av1_dec, 0, sizeof(*av1_dec));
>> +
>> +    av1_dec->global_model.cpu = dma_alloc_coherent(vpu->dev, 
>> GLOBAL_MODEL_SIZE,
>> + &av1_dec->global_model.dma,
>> +                               GFP_KERNEL);
>> +    if (!av1_dec->global_model.cpu)
>> +        return -ENOMEM;
>> +    av1_dec->global_model.size = GLOBAL_MODEL_SIZE;
>> +
>> +    av1_dec->tile_info.cpu = dma_alloc_coherent(vpu->dev, 
>> AV1_TILE_INFO_SIZE,
>> +                            &av1_dec->tile_info.dma,
>> +                            GFP_KERNEL);
>> +    if (!av1_dec->tile_info.cpu)
>> +        return -ENOMEM;
>> +    av1_dec->tile_info.size = AV1_TILE_INFO_SIZE;
>> +
>> +    av1_dec->film_grain.cpu = dma_alloc_coherent(vpu->dev,
>> +                             ALIGN(sizeof(struct 
>> hantro_av1_film_grain),
>> +                               2048),
>> +                             &av1_dec->film_grain.dma,
>> +                             GFP_KERNEL);
>> +    if (!av1_dec->film_grain.cpu)
>> +        return -ENOMEM;
>> +    av1_dec->film_grain.size = ALIGN(sizeof(struct 
>> hantro_av1_film_grain), 2048);
>> +
>> +    av1_dec->prob_tbl.cpu = dma_alloc_coherent(vpu->dev,
>> +                           ALIGN(sizeof(struct av1cdfs), 2048),
>> +                           &av1_dec->prob_tbl.dma,
>> +                           GFP_KERNEL);
>> +    if (!av1_dec->prob_tbl.cpu)
>> +        return -ENOMEM;
>> +    av1_dec->prob_tbl.size = ALIGN(sizeof(struct av1cdfs), 2048);
>> +
>> +    av1_dec->prob_tbl_out.cpu = dma_alloc_coherent(vpu->dev,
>> +                               ALIGN(sizeof(struct av1cdfs), 2048),
>> + &av1_dec->prob_tbl_out.dma,
>> +                               GFP_KERNEL);
>> +    if (!av1_dec->prob_tbl_out.cpu)
>> +        return -ENOMEM;
>> +    av1_dec->prob_tbl_out.size = ALIGN(sizeof(struct av1cdfs), 2048);
>> +    av1_dec->cdfs = &av1_dec->default_cdfs;
>> +    av1_dec->cdfs_ndvc = &av1_dec->default_cdfs_ndvc;
>> +
>> +    hantro_av1_set_default_cdfs(av1_dec->cdfs, av1_dec->cdfs_ndvc);
>> +
>> +    av1_dec->tile_buf.cpu = dma_alloc_coherent(vpu->dev,
>> +                           AV1_TILE_SIZE,
>> +                           &av1_dec->tile_buf.dma,
>> +                           GFP_KERNEL);
>> +    if (!av1_dec->tile_buf.cpu)
>> +        return -ENOMEM;
>> +    av1_dec->tile_buf.size = AV1_TILE_SIZE;
>> +
>> +    return 0;
>> +}
>> +
>> +int hantro_av1_prepare_run(struct hantro_ctx *ctx)
>> +{
>> +    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> +    struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>> +
>> +    ctrls->sequence = hantro_get_ctrl(ctx, 
>> V4L2_CID_STATELESS_AV1_SEQUENCE);
>> +    if (WARN_ON(!ctrls->sequence))
>> +        return -EINVAL;
>> +
>> +    ctrls->tile_group_entry =
>> +        hantro_get_ctrl(ctx, V4L2_CID_STATELESS_AV1_TILE_GROUP_ENTRY);
>> +    if (WARN_ON(!ctrls->tile_group_entry))
>> +        return -EINVAL;
>> +
>> +    ctrls->frame = hantro_get_ctrl(ctx, V4L2_CID_STATELESS_AV1_FRAME);
>> +    if (WARN_ON(!ctrls->frame))
>> +        return -EINVAL;
>> +
>> +    ctrls->film_grain =
>> +        hantro_get_ctrl(ctx, V4L2_CID_STATELESS_AV1_FILM_GRAIN);
>> +
>> +    return hantro_av1_tiles_reallocate(ctx);
>> +}
>> +
>> +static int hantro_av1_get_msb(u32 n)
>> +{
>> +    if (n == 0)
>> +        return 0;
>> +    return 31 ^ __builtin_clz(n);
>> +}
>> +
>> +static short hantro_av1_resolve_divisor_32(u32 d, short *shift)
>> +{
>> +    int f;
>> +    u64 e;
>> +
>> +    *shift = hantro_av1_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 hantro_av1_get_shear_params(const u32 *params, s64 *alpha,
>> +                    s64 *beta, s64 *gamma, s64 *delta)
>> +{
>> +    const int *mat = params;
>> +    short shift;
>> +    short y;
>> +    long long gv, dv;
>> +
>> +    if (mat[2] <= 0)
>> +        return;
>> +
>> +    *alpha = clamp_val(mat[2] - (1 << WARPEDMODEL_PREC_BITS), 
>> S16_MIN, S16_MAX);
>> +    *beta = clamp_val(mat[3], S16_MIN, S16_MAX);
>> +
>> +    y = hantro_av1_resolve_divisor_32(abs(mat[2]), &shift) * (mat[2] 
>> < 0 ? -1 : 1);
>> +
>> +    gv = ((long long)mat[4] * (1 << WARPEDMODEL_PREC_BITS)) * y;
>> +
>> +    *gamma = clamp_val((int)AV1_DIV_ROUND_UP_POW2_SIGNED(gv, shift), 
>> S16_MIN, S16_MAX);
>> +
>> +    dv = ((long long)mat[3] * mat[4]) * y;
>> +    *delta = clamp_val(mat[5] -
>> +        (int)AV1_DIV_ROUND_UP_POW2_SIGNED(dv, shift) - (1 << 
>> WARPEDMODEL_PREC_BITS),
>> +        S16_MIN, S16_MAX);
>> +
>> +    *alpha = AV1_DIV_ROUND_UP_POW2_SIGNED(*alpha, 
>> WARP_PARAM_REDUCE_BITS)
>> +         * (1 << WARP_PARAM_REDUCE_BITS);
>> +    *beta = AV1_DIV_ROUND_UP_POW2_SIGNED(*beta, WARP_PARAM_REDUCE_BITS)
>> +        * (1 << WARP_PARAM_REDUCE_BITS);
>> +    *gamma = AV1_DIV_ROUND_UP_POW2_SIGNED(*gamma, 
>> WARP_PARAM_REDUCE_BITS)
>> +         * (1 << WARP_PARAM_REDUCE_BITS);
>> +    *delta = AV1_DIV_ROUND_UP_POW2_SIGNED(*delta, 
>> WARP_PARAM_REDUCE_BITS)
>> +        * (1 << WARP_PARAM_REDUCE_BITS);
>> +}
>> +
>> +void hantro_av1_set_global_model(struct hantro_ctx *ctx)
>> +{
>> +    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> +    struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>> +    const struct v4l2_ctrl_av1_frame *frame = ctrls->frame;
>> +    const struct v4l2_av1_global_motion *gm = &frame->global_motion;
>> +    u8 *dst = av1_dec->global_model.cpu;
>> +    int ref_frame, i;
>> +
>> +    memset(dst, 0, GLOBAL_MODEL_SIZE);
>> +    for (ref_frame = 0; ref_frame < V4L2_AV1_REFS_PER_FRAME; 
>> ++ref_frame) {
>> +        s64 alpha = 0, beta = 0, gamma = 0, delta = 0;
>> +
>> +        for (i = 0; i < 6; ++i) {
>> +            if (i == 2)
>> +                *(s32 *)dst =
>> +                    gm->params[V4L2_AV1_REF_LAST_FRAME + ref_frame][3];
>> +            else if (i == 3)
>> +                *(s32 *)dst =
>> +                    gm->params[V4L2_AV1_REF_LAST_FRAME + ref_frame][2];
>> +            else
>> +                *(s32 *)dst =
>> +                    gm->params[V4L2_AV1_REF_LAST_FRAME + ref_frame][i];
>> +            dst += 4;
>> +        }
>> +
>> +        if (gm->type[V4L2_AV1_REF_LAST_FRAME + ref_frame] <= 
>> V4L2_AV1_WARP_MODEL_AFFINE)
>> + hantro_av1_get_shear_params(&gm->params[V4L2_AV1_REF_LAST_FRAME + 
>> ref_frame][0],
>> +                            &alpha, &beta, &gamma, &delta);
>> +
>> +        *(s16 *)dst = alpha;
>> +        dst += 2;
>> +        *(s16 *)dst = beta;
>> +        dst += 2;
>> +        *(s16 *)dst = gamma;
>> +        dst += 2;
>> +        *(s16 *)dst = delta;
>> +        dst += 2;
>> +    }
>> +}
>> +
>> +int hantro_av1_tile_log2(int target)
>> +{
>> +    int k;
>> +
>> +    /*
>> +     * returns the smallest value for k such that 1 << k is greater
>> +     * than or equal to target
>> +     */
>> +    for (k = 0; (1 << k) < target; k++)
>> +        ;
>> +
>> +    return k;
>> +}
>> +
>> +int hantro_av1_get_dist(struct hantro_ctx *ctx, int a, int b)
>> +{
>> +    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> +    struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>> +    int bits = ctrls->sequence->order_hint_bits - 1;
>> +    int diff, m;
>> +
>> +    if (!ctrls->sequence->order_hint_bits)
>> +        return 0;
>> +
>> +    diff = a - b;
>> +    m = 1 << bits;
>> +    diff = (diff & (m - 1)) - (diff & m);
>> +
>> +    return diff;
>> +}
>> +
>> +void hantro_av1_set_frame_sign_bias(struct hantro_ctx *ctx)
>> +{
>> +    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> +    struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>> +    const struct v4l2_ctrl_av1_frame *frame = ctrls->frame;
>> +    const struct v4l2_ctrl_av1_sequence *sequence = ctrls->sequence;
>> +    int i;
>> +
>> +    if (!sequence->order_hint_bits || IS_INTRA(frame->frame_type)) {
>> +        for (i = 0; i < V4L2_AV1_TOTAL_REFS_PER_FRAME; i++)
>> +            av1_dec->ref_frame_sign_bias[i] = 0;
>> +
>> +        return;
>> +    }
>> +    // Identify the nearest forward and backward references.
>> +    for (i = 0; i < V4L2_AV1_TOTAL_REFS_PER_FRAME - 1; i++) {
>> +        if (hantro_av1_get_frame_index(ctx, i) >= 0) {
>> +            int rel_off =
>> +                hantro_av1_get_dist(ctx,
>> +                        hantro_av1_get_order_hint(ctx, i),
>> +                        frame->order_hint);
>> +            av1_dec->ref_frame_sign_bias[i + 1] = (rel_off <= 0) ? 0 
>> : 1;
>> +        }
>> +    }
>> +}
>> +
>> +void hantro_av1_init_scaling_function(const u8 *values, const u8 
>> *scaling,
>> +                      u8 num_points, u8 *scaling_lut)
>> +{
>> +    int i, point;
>> +
>> +    if (num_points == 0) {
>> +        memset(scaling_lut, 0, 256);
>> +        return;
>> +    }
>> +
>> +    for (point = 0; point < num_points - 1; point++) {
>> +        int x;
>> +        s32 delta_y = scaling[point + 1] - scaling[point];
>> +        s32 delta_x = values[point + 1] - values[point];
>> +        s64 delta =
>> +            delta_x ? delta_y * ((65536 + (delta_x >> 1)) /
>> +                     delta_x) : 0;
>> +
>> +        for (x = 0; x < delta_x; x++) {
>> +            scaling_lut[values[point] + x] =
>> +                scaling[point] +
>> +                (s32)((x * delta + 32768) >> 16);
>> +        }
>> +    }
>> +
>> +    for (i = values[num_points - 1]; i < 256; i++)
>> +        scaling_lut[i] = scaling[num_points - 1];
>> +}
>> +
>> +void hantro_av1_set_tile_info(struct hantro_ctx *ctx)
>> +{
>> +    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> +    struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>> +    const struct v4l2_av1_tile_info *tile_info = 
>> &ctrls->frame->tile_info;
>> +    const struct v4l2_ctrl_av1_tile_group_entry *group_entry =
>> +        ctrls->tile_group_entry;
>> +    u8 *dst = av1_dec->tile_info.cpu;
>> +    int tile0, tile1;
>> +
>> +    memset(dst, 0, av1_dec->tile_info.size);
>> +
>> +    for (tile0 = 0; tile0 < tile_info->tile_cols; tile0++) {
>> +        for (tile1 = 0; tile1 < tile_info->tile_rows; tile1++) {
>> +            int tile_id = tile1 * tile_info->tile_cols + tile0;
>> +            u32 start, end;
>> +            u32 y0 =
>> +                tile_info->height_in_sbs_minus_1[tile1] + 1;
>> +            u32 x0 = tile_info->width_in_sbs_minus_1[tile0] + 1;
>> +
>> +            /* tile size in SB units (width,height) */
>> +            *dst++ = x0;
>> +            *dst++ = 0;
>> +            *dst++ = 0;
>> +            *dst++ = 0;
>> +            *dst++ = y0;
>> +            *dst++ = 0;
>> +            *dst++ = 0;
>> +            *dst++ = 0;
>> +
>> +            /* tile start position */
>> +            start = group_entry[tile_id].tile_offset - 
>> group_entry[0].tile_offset;
>> +            *dst++ = start & 255;
>> +            *dst++ = (start >> 8) & 255;
>> +            *dst++ = (start >> 16) & 255;
>> +            *dst++ = (start >> 24) & 255;
>> +
>> +            /* number of bytes in tile data */
>> +            end = start + group_entry[tile_id].tile_size;
>> +            *dst++ = end & 255;
>> +            *dst++ = (end >> 8) & 255;
>> +            *dst++ = (end >> 16) & 255;
>> +            *dst++ = (end >> 24) & 255;
>> +        }
>> +    }
>> +}
>> +
>> +bool hantro_av1_is_lossless(struct hantro_ctx *ctx)
>> +{
>> +    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> +    struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>> +    const struct v4l2_ctrl_av1_frame *frame = ctrls->frame;
>> +    const struct v4l2_av1_segmentation *segmentation = 
>> &frame->segmentation;
>> +    const struct v4l2_av1_quantization *quantization = 
>> &frame->quantization;
>> +    int i;
>> +
>> +    for (i = 0; i < V4L2_AV1_MAX_SEGMENTS; i++) {
>> +        int qindex = quantization->base_q_idx;
>> +
>> +        if (segmentation->feature_enabled[i] &
>> + V4L2_AV1_SEGMENT_FEATURE_ENABLED(V4L2_AV1_SEG_LVL_ALT_Q)) {
>> +            qindex += 
>> segmentation->feature_data[i][V4L2_AV1_SEG_LVL_ALT_Q];
>> +        }
>> +        qindex = clamp(qindex, 0, 255);
>> +
>> +        if (qindex ||
>> +            quantization->delta_q_y_dc ||
>> +            quantization->delta_q_u_dc ||
>> +            quantization->delta_q_u_ac ||
>> +            quantization->delta_q_v_dc ||
>> +            quantization->delta_q_v_ac)
>> +            return false;
>> +    }
>> +
>> +    return true;
>> +}
>> +
>> +void hantro_av1_update_prob(struct hantro_ctx *ctx)
>> +{
>> +    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> +    struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>> +    const struct v4l2_ctrl_av1_frame *frame = ctrls->frame;
>> +    bool frame_is_intra = IS_INTRA(frame->frame_type);
>> +    struct av1cdfs *out_cdfs = (struct av1cdfs 
>> *)av1_dec->prob_tbl_out.cpu;
>> +    int i;
>> +
>> +    if (frame->flags & 
>> V4L2_AV1_FRAME_FLAG_DISABLE_FRAME_END_UPDATE_CDF)
>> +        return;
>> +
>> +    for (i = 0; i < NUM_REF_FRAMES; i++) {
>> +        if (frame->refresh_frame_flags & BIT(i)) {
>> +            struct mvcdfs stored_mv_cdf;
>> +
>> +            hantro_av1_get_cdfs(ctx, i);
>> +            stored_mv_cdf = av1_dec->cdfs->mv_cdf;
>> +            *av1_dec->cdfs = *out_cdfs;
>> +            if (frame_is_intra) {
>> +                av1_dec->cdfs->mv_cdf = stored_mv_cdf;
>> +                *av1_dec->cdfs_ndvc = out_cdfs->mv_cdf;
>> +            }
>> +            hantro_av1_store_cdfs(ctx, frame->refresh_frame_flags);
>> +            break;
>> +        }
>> +    }
>> +}
>> +
>> +void hantro_av1_set_prob(struct hantro_ctx *ctx)
>> +{
>> +    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> +    struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>> +    const struct v4l2_ctrl_av1_frame *frame = ctrls->frame;
>> +    const struct v4l2_av1_quantization *quantization = 
>> &frame->quantization;
>> +    bool error_resilient_mode =
>> +        !!(frame->flags & V4L2_AV1_FRAME_FLAG_ERROR_RESILIENT_MODE);
>> +    bool frame_is_intra = IS_INTRA(frame->frame_type);
>> +
>> +    if (error_resilient_mode || frame_is_intra ||
>> +        frame->primary_ref_frame == AV1_PRIMARY_REF_NONE) {
>> +        av1_dec->cdfs = &av1_dec->default_cdfs;
>> +        av1_dec->cdfs_ndvc = &av1_dec->default_cdfs_ndvc;
>> + hantro_av1_default_coeff_probs(quantization->base_q_idx,
>> +                           av1_dec->cdfs);
>> +    } else {
>> +        if (frame->primary_ref_frame < 0 ||
>> +            frame->primary_ref_frame >= V4L2_AV1_REFS_PER_FRAME)
>> +            return;
>> +
>> +        hantro_av1_get_cdfs(ctx, 
>> frame->ref_frame_idx[frame->primary_ref_frame]);
>> +    }
>> +    hantro_av1_store_cdfs(ctx, frame->refresh_frame_flags);
>> +
>> +    memcpy(av1_dec->prob_tbl.cpu, av1_dec->cdfs, sizeof(struct 
>> av1cdfs));
>> +
>> +    if (frame_is_intra) {
>> +        int mv_offset = offsetof(struct av1cdfs, mv_cdf);
>> +        /* Overwrite MV context area with intrabc MV context */
>> +        memcpy(av1_dec->prob_tbl.cpu + mv_offset, av1_dec->cdfs_ndvc,
>> +               sizeof(struct mvcdfs));
>> +    }
>> +}
>> diff --git a/drivers/media/platform/verisilicon/hantro_av1.h 
>> b/drivers/media/platform/verisilicon/hantro_av1.h
>> new file mode 100644
>> index 000000000000..d91c814eefcb
>> --- /dev/null
>> +++ b/drivers/media/platform/verisilicon/hantro_av1.h
>> @@ -0,0 +1,64 @@
>> +/* SPDX-License-Identifier: GPL-2.0-only */
>> +
>> +#ifndef _HANTRO_AV1_H_
>> +#define _HANTRO_AV1_H_
>> +
>> +#define AV1_PRIMARY_REF_NONE    7
>> +#define AV1_REF_SCALE_SHIFT    14
>> +#define MAX_FRAME_DISTANCE    31
>> +#define AV1DEC_MAX_PIC_BUFFERS    24
>> +
>> +#define SCALE_NUMERATOR 8
>> +#define SUPERRES_SCALE_DENOMINATOR_MIN (SCALE_NUMERATOR + 1)
>> +
>> +#define RS_SCALE_SUBPEL_BITS 14
>> +#define RS_SCALE_SUBPEL_MASK ((1 << RS_SCALE_SUBPEL_BITS) - 1)
>> +#define RS_SUBPEL_BITS 6
>> +#define RS_SUBPEL_MASK ((1 << RS_SUBPEL_BITS) - 1)
>> +#define RS_SCALE_EXTRA_BITS (RS_SCALE_SUBPEL_BITS - RS_SUBPEL_BITS)
>> +#define RS_SCALE_EXTRA_OFF (1 << (RS_SCALE_EXTRA_BITS - 1))
>> +
>> +/*
>> + * These 3 values aren't defined enum v4l2_av1_segment_feature because
>> + * they are not part of the specification
>> + */
>> +#define V4L2_AV1_SEG_LVL_ALT_LF_Y_H    2
>> +#define V4L2_AV1_SEG_LVL_ALT_LF_U    3
>> +#define V4L2_AV1_SEG_LVL_ALT_LF_V    4
>> +
>> +#define IS_INTRA(type) ((type == V4L2_AV1_KEY_FRAME) || (type == 
>> V4L2_AV1_INTRA_ONLY_FRAME))
>> +
>> +#define LST_BUF_IDX (V4L2_AV1_REF_LAST_FRAME - V4L2_AV1_REF_LAST_FRAME)
>> +#define LST2_BUF_IDX (V4L2_AV1_REF_LAST2_FRAME - 
>> V4L2_AV1_REF_LAST_FRAME)
>> +#define LST3_BUF_IDX (V4L2_AV1_REF_LAST3_FRAME - 
>> V4L2_AV1_REF_LAST_FRAME)
>> +#define GLD_BUF_IDX (V4L2_AV1_REF_GOLDEN_FRAME - 
>> V4L2_AV1_REF_LAST_FRAME)
>> +#define BWD_BUF_IDX (V4L2_AV1_REF_BWDREF_FRAME - 
>> V4L2_AV1_REF_LAST_FRAME)
>> +#define ALT2_BUF_IDX (V4L2_AV1_REF_ALTREF2_FRAME - 
>> V4L2_AV1_REF_LAST_FRAME)
>> +#define ALT_BUF_IDX (V4L2_AV1_REF_ALTREF_FRAME - 
>> V4L2_AV1_REF_LAST_FRAME)
>> +
>> +#define AV1_INVALID_IDX        -1
>> +
>> +int hantro_av1_get_frame_index(struct hantro_ctx *ctx, int ref);
>> +int hantro_av1_get_order_hint(struct hantro_ctx *ctx, int ref);
>> +int hantro_av1_frame_ref(struct hantro_ctx *ctx, u64 timestamp);
>> +void hantro_av1_clean_refs(struct hantro_ctx *ctx);
>> +size_t hantro_av1_luma_size(struct hantro_ctx *ctx);
>> +size_t hantro_av1_chroma_size(struct hantro_ctx *ctx);
>> +void hantro_av1_exit(struct hantro_ctx *ctx);
>> +int hantro_av1_init(struct hantro_ctx *ctx);
>> +int hantro_av1_prepare_run(struct hantro_ctx *ctx);
>> +void hantro_av1_set_global_model(struct hantro_ctx *ctx);
>> +int hantro_av1_tile_log2(int target);
>> +int hantro_av1_get_dist(struct hantro_ctx *ctx, int a, int b);
>> +void hantro_av1_set_frame_sign_bias(struct hantro_ctx *ctx);
>> +void hantro_av1_init_scaling_function(const u8 *values, const u8 
>> *scaling,
>> +                      u8 num_points, u8 *scaling_lut);
>> +void hantro_av1_set_tile_info(struct hantro_ctx *ctx);
>> +bool hantro_av1_is_lossless(struct hantro_ctx *ctx);
>> +void hantro_av1_update_prob(struct hantro_ctx *ctx);
>> +void hantro_av1_set_prob(struct hantro_ctx *ctx);
>> +
>> +int hantro_av1_get_hardware_mcomp_filt_type(int interpolation_filter);
>> +int hantro_av1_get_hardware_tx_mode(enum v4l2_av1_tx_mode tx_mode);
>> +
>> +#endif
>> diff --git 
>> a/drivers/media/platform/verisilicon/rockchip_av1_entropymode.c 
>> b/drivers/media/platform/verisilicon/hantro_av1_entropymode.c
>> similarity index 99%
>> rename from 
>> drivers/media/platform/verisilicon/rockchip_av1_entropymode.c
>> rename to drivers/media/platform/verisilicon/hantro_av1_entropymode.c
>> index b1ae72ad675e..4f7bfec73668 100644
>> --- a/drivers/media/platform/verisilicon/rockchip_av1_entropymode.c
>> +++ b/drivers/media/platform/verisilicon/hantro_av1_entropymode.c
>> @@ -11,7 +11,7 @@
>>    */
>>     #include "hantro.h"
>> -#include "rockchip_av1_entropymode.h"
>> +#include "hantro_av1_entropymode.h"
>>     #define AOM_ICDF ICDF
>>   #define AOM_CDF2(a0) AOM_ICDF(a0)
>> @@ -4195,7 +4195,7 @@ static const u16 default_bits_cdf[][10] = {
>>       }
>>   };
>>   -static int rockchip_av1_get_q_ctx(int q)
>> +static int hantro_av1_get_q_ctx(int q)
>>   {
>>       if (q <= 20)
>>           return 0;
>> @@ -4206,10 +4206,10 @@ static int rockchip_av1_get_q_ctx(int q)
>>       return 3;
>>   }
>>   -void rockchip_av1_default_coeff_probs(u32 base_qindex, void *ptr)
>> +void hantro_av1_default_coeff_probs(u32 base_qindex, void *ptr)
>>   {
>>       struct av1cdfs *cdfs = (struct av1cdfs *)ptr;
>> -    const int index = rockchip_av1_get_q_ctx(base_qindex);
>> +    const int index = hantro_av1_get_q_ctx(base_qindex);
>>         memcpy(cdfs->txb_skip_cdf, av1_default_txb_skip_cdfs[index],
>>              sizeof(av1_default_txb_skip_cdfs[0]));
>> @@ -4240,8 +4240,8 @@ void rockchip_av1_default_coeff_probs(u32 
>> base_qindex, void *ptr)
>>              sizeof(av1_default_eob_multi1024_cdfs[0]));
>>   }
>>   -void rockchip_av1_set_default_cdfs(struct av1cdfs *cdfs,
>> -                   struct mvcdfs *cdfs_ndvc)
>> +void hantro_av1_set_default_cdfs(struct av1cdfs *cdfs,
>> +                 struct mvcdfs *cdfs_ndvc)
>>   {
>>       memcpy(cdfs->partition_cdf, default_partition_cdf,
>>              sizeof(cdfs->partition_cdf));
>> @@ -4398,7 +4398,7 @@ void rockchip_av1_set_default_cdfs(struct 
>> av1cdfs *cdfs,
>>              sizeof(cdfs->compound_idx_cdf));
>>   }
>>   -void rockchip_av1_get_cdfs(struct hantro_ctx *ctx, u32 ref_idx)
>> +void hantro_av1_get_cdfs(struct hantro_ctx *ctx, u32 ref_idx)
>>   {
>>       struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>>   @@ -4406,8 +4406,8 @@ void rockchip_av1_get_cdfs(struct hantro_ctx 
>> *ctx, u32 ref_idx)
>>       av1_dec->cdfs_ndvc = &av1_dec->cdfs_last_ndvc[ref_idx];
>>   }
>>   -void rockchip_av1_store_cdfs(struct hantro_ctx *ctx,
>> -                 u32 refresh_frame_flags)
>> +void hantro_av1_store_cdfs(struct hantro_ctx *ctx,
>> +               u32 refresh_frame_flags)
>>   {
>>       struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>>       int i;
>> diff --git 
>> a/drivers/media/platform/verisilicon/rockchip_av1_entropymode.h 
>> b/drivers/media/platform/verisilicon/hantro_av1_entropymode.h
>> similarity index 95%
>> rename from 
>> drivers/media/platform/verisilicon/rockchip_av1_entropymode.h
>> rename to drivers/media/platform/verisilicon/hantro_av1_entropymode.h
>> index bbf8424c7d2c..abbc660ecce3 100644
>> --- a/drivers/media/platform/verisilicon/rockchip_av1_entropymode.h
>> +++ b/drivers/media/platform/verisilicon/hantro_av1_entropymode.h
>> @@ -1,7 +1,7 @@
>>   /* SPDX-License-Identifier: GPL-2.0-only */
>>   -#ifndef _ROCKCHIP_AV1_ENTROPYMODE_H_
>> -#define _ROCKCHIP_AV1_ENTROPYMODE_H_
>> +#ifndef _HANTRO_AV1_ENTROPYMODE_H_
>> +#define _HANTRO_AV1_ENTROPYMODE_H_
>>     #include <linux/types.h>
>>   @@ -262,11 +262,11 @@ struct av1cdfs {
>>       u16 dummy3[16];
>>   };
>>   -void rockchip_av1_store_cdfs(struct hantro_ctx *ctx,
>> -                 u32 refresh_frame_flags);
>> -void rockchip_av1_get_cdfs(struct hantro_ctx *ctx, u32 ref_idx);
>> -void rockchip_av1_set_default_cdfs(struct av1cdfs *cdfs,
>> -                   struct mvcdfs *cdfs_ndvc);
>> -void rockchip_av1_default_coeff_probs(u32 base_qindex, void *ptr);
>> +void hantro_av1_store_cdfs(struct hantro_ctx *ctx,
>> +               u32 refresh_frame_flags);
>> +void hantro_av1_get_cdfs(struct hantro_ctx *ctx, u32 ref_idx);
>> +void hantro_av1_set_default_cdfs(struct av1cdfs *cdfs,
>> +                 struct mvcdfs *cdfs_ndvc);
>> +void hantro_av1_default_coeff_probs(u32 base_qindex, void *ptr);
>>   -#endif /* _ROCKCHIP_AV1_ENTROPYMODE_H_ */
>> +#endif /* _HANTRO_AV1_ENTROPYMODE_H_ */
>> diff --git 
>> a/drivers/media/platform/verisilicon/rockchip_av1_filmgrain.c 
>> b/drivers/media/platform/verisilicon/hantro_av1_filmgrain.c
>> similarity index 85%
>> rename from drivers/media/platform/verisilicon/rockchip_av1_filmgrain.c
>> rename to drivers/media/platform/verisilicon/hantro_av1_filmgrain.c
>> index f64dea797eff..a1fc40e0218a 100644
>> --- a/drivers/media/platform/verisilicon/rockchip_av1_filmgrain.c
>> +++ b/drivers/media/platform/verisilicon/hantro_av1_filmgrain.c
>> @@ -1,6 +1,6 @@
>>   // SPDX-License-Identifier: GPL-2.0-only OR Apache-2.0
>>   -#include "rockchip_av1_filmgrain.h"
>> +#include "hantro_av1_filmgrain.h"
>>     static const s32 gaussian_sequence[2048] = {
>>       56, 568, -180, 172, 124, -84, 172, -64, -900, 24, 820,
>> @@ -204,8 +204,8 @@ static inline s32 round_power_of_two(const s32 
>> val, s32 n)
>>       return (val + a) >> n;
>>   }
>>   -static void rockchip_av1_init_random_generator(u8 luma_num, u16 seed,
>> -                           u16 *random_register)
>> +static void hantro_av1_init_random_generator(u8 luma_num, u16 seed,
>> +                         u16 *random_register)
>>   {
>>       u16 random_reg = seed;
>>   @@ -214,7 +214,7 @@ static void 
>> rockchip_av1_init_random_generator(u8 luma_num, u16 seed,
>>       *random_register = random_reg;
>>   }
>>   -static inline void rockchip_av1_update_random_register(u16 
>> *random_register)
>> +static inline void hantro_av1_update_random_register(u16 
>> *random_register)
>>   {
>>       u16 bit;
>>       u16 random_reg = *random_register;
>> @@ -224,21 +224,20 @@ static inline void 
>> rockchip_av1_update_random_register(u16 *random_register)
>>       *random_register = (random_reg >> 1) | (bit << 15);
>>   }
>>   -static inline s32 rockchip_av1_get_random_number(u16 random_register)
>> +static inline s32 hantro_av1_get_random_number(u16 random_register)
>>   {
>>       return (random_register >> 5) & ((1 << 11) - 1);
>>   }
>>   -void rockchip_av1_generate_luma_grain_block(s32 
>> (*luma_grain_block)[73][82],
>> -                        s32 bitdepth,
>> -                        u8 num_y_points,
>> -                        s32 grain_scale_shift,
>> -                        s32 ar_coeff_lag,
>> -                        s32 (*ar_coeffs_y)[24],
>> -                        s32 ar_coeff_shift,
>> -                        s32 grain_min,
>> -                        s32 grain_max,
>> -                        u16 random_seed)
>> +void hantro_av1_generate_luma_grain_block(struct 
>> hantro_av1_coeffs_grain_block *coeffs,
>> +                      s32 bitdepth,
>> +                      u8 num_y_points,
>> +                      s32 grain_scale_shift,
>> +                      s32 ar_coeff_lag,
>> +                      s32 ar_coeff_shift,
>> +                      s32 grain_min,
>> +                      s32 grain_max,
>> +                      u16 random_seed)
>>   {
>>       s32 gauss_sec_shift = 12 - bitdepth + grain_scale_shift;
>>       u16 grain_random_register = random_seed;
>> @@ -247,15 +246,15 @@ void rockchip_av1_generate_luma_grain_block(s32 
>> (*luma_grain_block)[73][82],
>>       for (i = 0; i < 73; i++) {
>>           for (j = 0; j < 82; j++) {
>>               if (num_y_points > 0) {
>> -                rockchip_av1_update_random_register
>> +                hantro_av1_update_random_register
>>                       (&grain_random_register);
>> -                (*luma_grain_block)[i][j] =
>> +                coeffs->luma_grain_block[i][j] =
>>                       round_power_of_two(gaussian_sequence
>> -                         [rockchip_av1_get_random_number
>> +                         [hantro_av1_get_random_number
>>                             (grain_random_register)],
>>                            gauss_sec_shift);
>>               } else {
>> -                (*luma_grain_block)[i][j] = 0;
>> +                coeffs->luma_grain_block[i][j] = 0;
>>               }
>>           }
>>       }
>> @@ -272,72 +271,68 @@ void rockchip_av1_generate_luma_grain_block(s32 
>> (*luma_grain_block)[73][82],
>>                        deltacol <= ar_coeff_lag; deltacol++) {
>>                       if (deltarow == 0 && deltacol == 0)
>>                           break;
>> -                    wsum = wsum + (*ar_coeffs_y)[pos] *
>> -                        (*luma_grain_block)[i + deltarow][j + 
>> deltacol];
>> +                    wsum = wsum + coeffs->ar_coeffs_y[pos] *
>> +                        coeffs->luma_grain_block[i + deltarow][j + 
>> deltacol];
>>                       ++pos;
>>                   }
>>               }
>> -            (*luma_grain_block)[i][j] =
>> -                clamp((*luma_grain_block)[i][j] +
>> +            coeffs->luma_grain_block[i][j] =
>> +                clamp(coeffs->luma_grain_block[i][j] +
>>                         round_power_of_two(wsum, ar_coeff_shift),
>>                         grain_min, grain_max);
>>           }
>>   }
>>     // Calculate chroma grain noise once per frame
>> -void rockchip_av1_generate_chroma_grain_block(s32 
>> (*luma_grain_block)[73][82],
>> -                          s32 (*cb_grain_block)[38][44],
>> -                          s32 (*cr_grain_block)[38][44],
>> -                          s32 bitdepth,
>> -                          u8 num_y_points,
>> -                          u8 num_cb_points,
>> -                          u8 num_cr_points,
>> -                          s32 grain_scale_shift,
>> -                          s32 ar_coeff_lag,
>> -                          s32 (*ar_coeffs_cb)[25],
>> -                          s32 (*ar_coeffs_cr)[25],
>> -                          s32 ar_coeff_shift,
>> -                          s32 grain_min,
>> -                          s32 grain_max,
>> -                          u8 chroma_scaling_from_luma,
>> -                          u16 random_seed)
>> +void hantro_av1_generate_chroma_grain_block(struct 
>> hantro_av1_coeffs_grain_block *coeffs,
>> +                        s32 bitdepth,
>> +                        u8 num_y_points,
>> +                        u8 num_cb_points,
>> +                        u8 num_cr_points,
>> +                        s32 grain_scale_shift,
>> +                        s32 ar_coeff_lag,
>> +                        s32 ar_coeff_shift,
>> +                        s32 grain_min,
>> +                        s32 grain_max,
>> +                        u8 chroma_scaling_from_luma,
>> +                        u16 random_seed)
>>   {
>>       s32 gauss_sec_shift = 12 - bitdepth + grain_scale_shift;
>>       u16 grain_random_register = 0;
>>       s32 i, j;
>>   -    rockchip_av1_init_random_generator(7, random_seed,
>> -                       &grain_random_register);
>> +    hantro_av1_init_random_generator(7, random_seed,
>> +                     &grain_random_register);
>>       for (i = 0; i < 38; i++) {
>>           for (j = 0; j < 44; j++) {
>>               if (num_cb_points || chroma_scaling_from_luma) {
>> -                rockchip_av1_update_random_register
>> +                hantro_av1_update_random_register
>>                       (&grain_random_register);
>> -                (*cb_grain_block)[i][j] =
>> +                coeffs->cb_grain_block[i][j] =
>>                       round_power_of_two(gaussian_sequence
>> -                         [rockchip_av1_get_random_number
>> +                         [hantro_av1_get_random_number
>>                             (grain_random_register)],
>>                            gauss_sec_shift);
>>               } else {
>> -                (*cb_grain_block)[i][j] = 0;
>> +                coeffs->cb_grain_block[i][j] = 0;
>>               }
>>           }
>>       }
>>   -    rockchip_av1_init_random_generator(11, random_seed,
>> -                       &grain_random_register);
>> +    hantro_av1_init_random_generator(11, random_seed,
>> +                     &grain_random_register);
>>       for (i = 0; i < 38; i++) {
>>           for (j = 0; j < 44; j++) {
>>               if (num_cr_points || chroma_scaling_from_luma) {
>> -                rockchip_av1_update_random_register
>> +                hantro_av1_update_random_register
>>                       (&grain_random_register);
>> -                (*cr_grain_block)[i][j] =
>> +                coeffs->cr_grain_block[i][j] =
>>                       round_power_of_two(gaussian_sequence
>> -                         [rockchip_av1_get_random_number
>> +                         [hantro_av1_get_random_number
>>                             (grain_random_register)],
>>                            gauss_sec_shift);
>>               } else {
>> -                (*cr_grain_block)[i][j] = 0;
>> +                coeffs->cr_grain_block[i][j] = 0;
>>               }
>>           }
>>       }
>> @@ -355,12 +350,12 @@ void 
>> rockchip_av1_generate_chroma_grain_block(s32 
>> (*luma_grain_block)[73][82],
>>                        deltacol <= ar_coeff_lag; deltacol++) {
>>                       if (deltarow == 0 && deltacol == 0)
>>                           break;
>> -                    wsum_cb = wsum_cb + (*ar_coeffs_cb)[pos] *
>> -                        (*cb_grain_block)[i + deltarow][j + deltacol];
>> +                    wsum_cb = wsum_cb + coeffs->ar_coeffs_cb[pos] *
>> +                        coeffs->cb_grain_block[i + deltarow][j + 
>> deltacol];
>>                       wsum_cr =
>>                           wsum_cr +
>> -                        (*ar_coeffs_cr)[pos] *
>> -                        (*cr_grain_block)[i + deltarow][j + deltacol];
>> +                        coeffs->ar_coeffs_cr[pos] *
>> +                        coeffs->cr_grain_block[i + deltarow][j + 
>> deltacol];
>>                       ++pos;
>>                   }
>>               }
>> @@ -371,28 +366,28 @@ void 
>> rockchip_av1_generate_chroma_grain_block(s32 
>> (*luma_grain_block)[73][82],
>>                   s32 luma_coord_x = (j << 1) - 3;
>>                     av_luma +=
>> - (*luma_grain_block)[luma_coord_y][luma_coord_x];
>> + coeffs->luma_grain_block[luma_coord_y][luma_coord_x];
>>                   av_luma +=
>> - (*luma_grain_block)[luma_coord_y][luma_coord_x + 1];
>> + coeffs->luma_grain_block[luma_coord_y][luma_coord_x + 1];
>>                   av_luma +=
>> -                    (*luma_grain_block)[luma_coord_y + 
>> 1][luma_coord_x];
>> +                    coeffs->luma_grain_block[luma_coord_y + 
>> 1][luma_coord_x];
>>                   av_luma +=
>> -                    (*luma_grain_block)[(luma_coord_y + 
>> 1)][luma_coord_x + 1];
>> +                    coeffs->luma_grain_block[(luma_coord_y + 
>> 1)][luma_coord_x + 1];
>>                   av_luma = round_power_of_two(av_luma, 2);
>>   -                wsum_cb = wsum_cb + (*ar_coeffs_cb)[pos] * av_luma;
>> -                wsum_cr = wsum_cr + (*ar_coeffs_cr)[pos] * av_luma;
>> +                wsum_cb = wsum_cb + coeffs->ar_coeffs_cb[pos] * 
>> av_luma;
>> +                wsum_cr = wsum_cr + coeffs->ar_coeffs_cr[pos] * 
>> av_luma;
>>               }
>>                 if (num_cb_points || chroma_scaling_from_luma) {
>> -                (*cb_grain_block)[i][j] =
>> -                    clamp((*cb_grain_block)[i][j] +
>> +                coeffs->cb_grain_block[i][j] =
>> +                    clamp(coeffs->cb_grain_block[i][j] +
>>                         round_power_of_two(wsum_cb, ar_coeff_shift),
>>                         grain_min, grain_max);
>>               }
>>               if (num_cr_points || chroma_scaling_from_luma) {
>> -                (*cr_grain_block)[i][j] =
>> -                    clamp((*cr_grain_block)[i][j] +
>> +                coeffs->cr_grain_block[i][j] =
>> +                    clamp(coeffs->cr_grain_block[i][j] +
>>                         round_power_of_two(wsum_cr, ar_coeff_shift),
>>                         grain_min, grain_max);
>>               }
>> diff --git 
>> a/drivers/media/platform/verisilicon/hantro_av1_filmgrain.h 
>> b/drivers/media/platform/verisilicon/hantro_av1_filmgrain.h
>> new file mode 100644
>> index 000000000000..d80be7c011a0
>> --- /dev/null
>> +++ b/drivers/media/platform/verisilicon/hantro_av1_filmgrain.h
>> @@ -0,0 +1,48 @@
>> +/* SPDX-License-Identifier: GPL-2.0-only */
>> +
>> +#ifndef _HANTRO_AV1_FILMGRAIN_H_
>> +#define _HANTRO_AV1_FILMGRAIN_H_
>> +
>> +#include <linux/types.h>
>> +
>> +struct hantro_av1_film_grain {
>> +    u8 scaling_lut_y[256];
>> +    u8 scaling_lut_cb[256];
>> +    u8 scaling_lut_cr[256];
>> +    s16 cropped_luma_grain_block[4096];
>> +    s16 cropped_chroma_grain_block[1024 * 2];
>> +};
>> +
>> +struct hantro_av1_coeffs_grain_block {
>> +    s32 ar_coeffs_y[24];
>> +    s32 ar_coeffs_cb[25];
>> +    s32 ar_coeffs_cr[25];
>> +    s32 luma_grain_block[73][82];
>> +    s32 cb_grain_block[38][44];
>> +    s32 cr_grain_block[38][44];
>> +};
>> +
>> +void hantro_av1_generate_luma_grain_block(struct 
>> hantro_av1_coeffs_grain_block *coeffs,
>> +                      s32 bitdepth,
>> +                      u8 num_y_points,
>> +                      s32 grain_scale_shift,
>> +                      s32 ar_coeff_lag,
>> +                      s32 ar_coeff_shift,
>> +                      s32 grain_min,
>> +                      s32 grain_max,
>> +                      u16 random_seed);
>> +
>> +void hantro_av1_generate_chroma_grain_block(struct 
>> hantro_av1_coeffs_grain_block *coeffs,
>> +                        s32 bitdepth,
>> +                        u8 num_y_points,
>> +                        u8 num_cb_points,
>> +                        u8 num_cr_points,
>> +                        s32 grain_scale_shift,
>> +                        s32 ar_coeff_lag,
>> +                        s32 ar_coeff_shift,
>> +                        s32 grain_min,
>> +                        s32 grain_max,
>> +                        u8 chroma_scaling_from_luma,
>> +                        u16 random_seed);
>> +
>> +#endif
>> diff --git a/drivers/media/platform/verisilicon/hantro_hw.h 
>> b/drivers/media/platform/verisilicon/hantro_hw.h
>> index 5f2011529f02..d1d39d1df5d2 100644
>> --- a/drivers/media/platform/verisilicon/hantro_hw.h
>> +++ b/drivers/media/platform/verisilicon/hantro_hw.h
>> @@ -15,8 +15,8 @@
>>   #include <media/v4l2-vp9.h>
>>   #include <media/videobuf2-core.h>
>>   -#include "rockchip_av1_entropymode.h"
>> -#include "rockchip_av1_filmgrain.h"
>> +#include "hantro_av1_entropymode.h"
>> +#include "hantro_av1_filmgrain.h"
>>     #define DEC_8190_ALIGN_MASK    0x07U
>>   @@ -459,10 +459,7 @@ void hantro_hevc_ref_init(struct hantro_ctx 
>> *ctx);
>>   dma_addr_t hantro_hevc_get_ref_buf(struct hantro_ctx *ctx, s32 poc);
>>   int hantro_hevc_add_ref_buf(struct hantro_ctx *ctx, int poc, 
>> dma_addr_t addr);
>>   -int rockchip_vpu981_av1_dec_init(struct hantro_ctx *ctx);
>> -void rockchip_vpu981_av1_dec_exit(struct hantro_ctx *ctx);
>>   int rockchip_vpu981_av1_dec_run(struct hantro_ctx *ctx);
>> -void rockchip_vpu981_av1_dec_done(struct hantro_ctx *ctx);
>>     static inline unsigned short hantro_vp9_num_sbs(unsigned short 
>> dimension)
>>   {
>> diff --git 
>> a/drivers/media/platform/verisilicon/rockchip_av1_filmgrain.h 
>> b/drivers/media/platform/verisilicon/rockchip_av1_filmgrain.h
>> deleted file mode 100644
>> index 31a8b7920c31..000000000000
>> --- a/drivers/media/platform/verisilicon/rockchip_av1_filmgrain.h
>> +++ /dev/null
>> @@ -1,36 +0,0 @@
>> -/* SPDX-License-Identifier: GPL-2.0-only */
>> -
>> -#ifndef _ROCKCHIP_AV1_FILMGRAIN_H_
>> -#define _ROCKCHIP_AV1_FILMGRAIN_H_
>> -
>> -#include <linux/types.h>
>> -
>> -void rockchip_av1_generate_luma_grain_block(s32 
>> (*luma_grain_block)[73][82],
>> -                        s32 bitdepth,
>> -                        u8 num_y_points,
>> -                        s32 grain_scale_shift,
>> -                        s32 ar_coeff_lag,
>> -                        s32 (*ar_coeffs_y)[24],
>> -                        s32 ar_coeff_shift,
>> -                        s32 grain_min,
>> -                        s32 grain_max,
>> -                        u16 random_seed);
>> -
>> -void rockchip_av1_generate_chroma_grain_block(s32 
>> (*luma_grain_block)[73][82],
>> -                          s32 (*cb_grain_block)[38][44],
>> -                          s32 (*cr_grain_block)[38][44],
>> -                          s32 bitdepth,
>> -                          u8 num_y_points,
>> -                          u8 num_cb_points,
>> -                          u8 num_cr_points,
>> -                          s32 grain_scale_shift,
>> -                          s32 ar_coeff_lag,
>> -                          s32 (*ar_coeffs_cb)[25],
>> -                          s32 (*ar_coeffs_cr)[25],
>> -                          s32 ar_coeff_shift,
>> -                          s32 grain_min,
>> -                          s32 grain_max,
>> -                          u8 chroma_scaling_from_luma,
>> -                          u16 random_seed);
>> -
>> -#endif
>> diff --git 
>> a/drivers/media/platform/verisilicon/rockchip_vpu981_hw_av1_dec.c 
>> b/drivers/media/platform/verisilicon/rockchip_vpu981_hw_av1_dec.c
>> index e4e21ad37323..61e30338422d 100644
>> --- a/drivers/media/platform/verisilicon/rockchip_vpu981_hw_av1_dec.c
>> +++ b/drivers/media/platform/verisilicon/rockchip_vpu981_hw_av1_dec.c
>> @@ -7,622 +7,35 @@
>>     #include <media/v4l2-mem2mem.h>
>>   #include "hantro.h"
>> +#include "hantro_av1.h"
>>   #include "hantro_v4l2.h"
>>   #include "rockchip_vpu981_regs.h"
>>     #define AV1_DEC_MODE        17
>> -#define GM_GLOBAL_MODELS_PER_FRAME    7
>> -#define GLOBAL_MODEL_TOTAL_SIZE    (6 * 4 + 4 * 2)
>> -#define GLOBAL_MODEL_SIZE    ALIGN(GM_GLOBAL_MODELS_PER_FRAME * 
>> GLOBAL_MODEL_TOTAL_SIZE, 2048)
>> -#define AV1_MAX_TILES        128
>> -#define AV1_TILE_INFO_SIZE    (AV1_MAX_TILES * 16)
>> -#define AV1DEC_MAX_PIC_BUFFERS    24
>> -#define AV1_REF_SCALE_SHIFT    14
>> -#define AV1_INVALID_IDX        -1
>> -#define MAX_FRAME_DISTANCE    31
>> -#define AV1_PRIMARY_REF_NONE    7
>> -#define AV1_TILE_SIZE        ALIGN(32 * 128, 4096)
>> -/*
>> - * These 3 values aren't defined enum v4l2_av1_segment_feature because
>> - * they are not part of the specification
>> - */
>> -#define V4L2_AV1_SEG_LVL_ALT_LF_Y_H    2
>> -#define V4L2_AV1_SEG_LVL_ALT_LF_U    3
>> -#define V4L2_AV1_SEG_LVL_ALT_LF_V    4
>> -
>> -#define SUPERRES_SCALE_BITS 3
>> -#define SCALE_NUMERATOR 8
>> -#define SUPERRES_SCALE_DENOMINATOR_MIN (SCALE_NUMERATOR + 1)
>> -
>> -#define RS_SUBPEL_BITS 6
>> -#define RS_SUBPEL_MASK ((1 << RS_SUBPEL_BITS) - 1)
>> -#define RS_SCALE_SUBPEL_BITS 14
>> -#define RS_SCALE_SUBPEL_MASK ((1 << RS_SCALE_SUBPEL_BITS) - 1)
>> -#define RS_SCALE_EXTRA_BITS (RS_SCALE_SUBPEL_BITS - RS_SUBPEL_BITS)
>> -#define RS_SCALE_EXTRA_OFF (1 << (RS_SCALE_EXTRA_BITS - 1))
>> -
>> -#define IS_INTRA(type) ((type == V4L2_AV1_KEY_FRAME) || (type == 
>> V4L2_AV1_INTRA_ONLY_FRAME))
>> -
>> -#define LST_BUF_IDX (V4L2_AV1_REF_LAST_FRAME - V4L2_AV1_REF_LAST_FRAME)
>> -#define LST2_BUF_IDX (V4L2_AV1_REF_LAST2_FRAME - 
>> V4L2_AV1_REF_LAST_FRAME)
>> -#define LST3_BUF_IDX (V4L2_AV1_REF_LAST3_FRAME - 
>> V4L2_AV1_REF_LAST_FRAME)
>> -#define GLD_BUF_IDX (V4L2_AV1_REF_GOLDEN_FRAME - 
>> V4L2_AV1_REF_LAST_FRAME)
>> -#define BWD_BUF_IDX (V4L2_AV1_REF_BWDREF_FRAME - 
>> V4L2_AV1_REF_LAST_FRAME)
>> -#define ALT2_BUF_IDX (V4L2_AV1_REF_ALTREF2_FRAME - 
>> V4L2_AV1_REF_LAST_FRAME)
>> -#define ALT_BUF_IDX (V4L2_AV1_REF_ALTREF_FRAME - 
>> V4L2_AV1_REF_LAST_FRAME)
>> -
>> -#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 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_)));        \
>> -})
>> -
>> -enum rockchip_av1_tx_mode {
>> -    ROCKCHIP_AV1_TX_MODE_ONLY_4X4    = 0,
>> -    ROCKCHIP_AV1_TX_MODE_8X8    = 1,
>> -    ROCKCHIP_AV1_TX_MODE_16x16    = 2,
>> -    ROCKCHIP_AV1_TX_MODE_32x32    = 3,
>> -    ROCKCHIP_AV1_TX_MODE_SELECT    = 4,
>> -};
>> -
>> -struct rockchip_av1_film_grain {
>> -    u8 scaling_lut_y[256];
>> -    u8 scaling_lut_cb[256];
>> -    u8 scaling_lut_cr[256];
>> -    s16 cropped_luma_grain_block[4096];
>> -    s16 cropped_chroma_grain_block[1024 * 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,
>> -};
>> -
>> -static int rockchip_vpu981_get_frame_index(struct hantro_ctx *ctx, 
>> int ref)
>> -{
>> -    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> -    struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>> -    const struct v4l2_ctrl_av1_frame *frame = ctrls->frame;
>> -    u64 timestamp;
>> -    int i, idx = frame->ref_frame_idx[ref];
>> -
>> -    if (idx >= V4L2_AV1_TOTAL_REFS_PER_FRAME || idx < 0)
>> -        return AV1_INVALID_IDX;
>> -
>> -    timestamp = frame->reference_frame_ts[idx];
>> -    for (i = 0; i < AV1_MAX_FRAME_BUF_COUNT; i++) {
>> -        if (!av1_dec->frame_refs[i].used)
>> -            continue;
>> -        if (av1_dec->frame_refs[i].timestamp == timestamp)
>> -            return i;
>> -    }
>> -
>> -    return AV1_INVALID_IDX;
>> -}
>> -
>> -static int rockchip_vpu981_get_order_hint(struct hantro_ctx *ctx, 
>> int ref)
>> -{
>> -    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> -    int idx = rockchip_vpu981_get_frame_index(ctx, ref);
>> -
>> -    if (idx != AV1_INVALID_IDX)
>> -        return av1_dec->frame_refs[idx].order_hint;
>> -
>> -    return 0;
>> -}
>> -
>> -static int rockchip_vpu981_av1_dec_frame_ref(struct hantro_ctx *ctx,
>> -                         u64 timestamp)
>> -{
>> -    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> -    struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>> -    const struct v4l2_ctrl_av1_frame *frame = ctrls->frame;
>> -    int i;
>> -
>> -    for (i = 0; i < AV1_MAX_FRAME_BUF_COUNT; i++) {
>> -        int j;
>> -
>> -        if (av1_dec->frame_refs[i].used)
>> -            continue;
>> -
>> -        av1_dec->frame_refs[i].width = frame->frame_width_minus_1 + 1;
>> -        av1_dec->frame_refs[i].height = frame->frame_height_minus_1 
>> + 1;
>> -        av1_dec->frame_refs[i].mi_cols = 
>> DIV_ROUND_UP(frame->frame_width_minus_1 + 1, 8);
>> -        av1_dec->frame_refs[i].mi_rows = 
>> DIV_ROUND_UP(frame->frame_height_minus_1 + 1, 8);
>> -        av1_dec->frame_refs[i].timestamp = timestamp;
>> -        av1_dec->frame_refs[i].frame_type = frame->frame_type;
>> -        av1_dec->frame_refs[i].order_hint = frame->order_hint;
>> -        av1_dec->frame_refs[i].vb2_ref = hantro_get_dst_buf(ctx);
>> -
>> -        for (j = 0; j < V4L2_AV1_TOTAL_REFS_PER_FRAME; j++)
>> -            av1_dec->frame_refs[i].order_hints[j] = 
>> frame->order_hints[j];
>> -        av1_dec->frame_refs[i].used = true;
>> -        av1_dec->current_frame_index = i;
>> -
>> -        return i;
>> -    }
>> -
>> -    return AV1_INVALID_IDX;
>> -}
>> -
>> -static void rockchip_vpu981_av1_dec_frame_unref(struct hantro_ctx 
>> *ctx, int idx)
>> -{
>> -    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> -
>> -    if (idx >= 0)
>> -        av1_dec->frame_refs[idx].used = false;
>> -}
>> -
>> -static void rockchip_vpu981_av1_dec_clean_refs(struct hantro_ctx *ctx)
>> -{
>> -    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> -    struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>> -
>> -    int ref, idx;
>> -
>> -    for (idx = 0; idx < AV1_MAX_FRAME_BUF_COUNT; idx++) {
>> -        u64 timestamp = av1_dec->frame_refs[idx].timestamp;
>> -        bool used = false;
>> -
>> -        if (!av1_dec->frame_refs[idx].used)
>> -            continue;
>> -
>> -        for (ref = 0; ref < V4L2_AV1_TOTAL_REFS_PER_FRAME; ref++) {
>> -            if (ctrls->frame->reference_frame_ts[ref] == timestamp)
>> -                used = true;
>> -        }
>> -
>> -        if (!used)
>> -            rockchip_vpu981_av1_dec_frame_unref(ctx, idx);
>> -    }
>> -}
>> -
>> -static size_t rockchip_vpu981_av1_dec_luma_size(struct hantro_ctx *ctx)
>> -{
>> -    return ctx->dst_fmt.width * ctx->dst_fmt.height * ctx->bit_depth 
>> / 8;
>> -}
>> -
>> -static size_t rockchip_vpu981_av1_dec_chroma_size(struct hantro_ctx 
>> *ctx)
>> -{
>> -    size_t cr_offset = rockchip_vpu981_av1_dec_luma_size(ctx);
>> -
>> -    return ALIGN((cr_offset * 3) / 2, 64);
>> -}
>> -
>> -static void rockchip_vpu981_av1_dec_tiles_free(struct hantro_ctx *ctx)
>> -{
>> -    struct hantro_dev *vpu = ctx->dev;
>> -    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> -
>> -    if (av1_dec->db_data_col.cpu)
>> -        dma_free_coherent(vpu->dev, av1_dec->db_data_col.size,
>> -                  av1_dec->db_data_col.cpu,
>> -                  av1_dec->db_data_col.dma);
>> -    av1_dec->db_data_col.cpu = NULL;
>> -
>> -    if (av1_dec->db_ctrl_col.cpu)
>> -        dma_free_coherent(vpu->dev, av1_dec->db_ctrl_col.size,
>> -                  av1_dec->db_ctrl_col.cpu,
>> -                  av1_dec->db_ctrl_col.dma);
>> -    av1_dec->db_ctrl_col.cpu = NULL;
>> -
>> -    if (av1_dec->cdef_col.cpu)
>> -        dma_free_coherent(vpu->dev, av1_dec->cdef_col.size,
>> -                  av1_dec->cdef_col.cpu, av1_dec->cdef_col.dma);
>> -    av1_dec->cdef_col.cpu = NULL;
>> -
>> -    if (av1_dec->sr_col.cpu)
>> -        dma_free_coherent(vpu->dev, av1_dec->sr_col.size,
>> -                  av1_dec->sr_col.cpu, av1_dec->sr_col.dma);
>> -    av1_dec->sr_col.cpu = NULL;
>> -
>> -    if (av1_dec->lr_col.cpu)
>> -        dma_free_coherent(vpu->dev, av1_dec->lr_col.size,
>> -                  av1_dec->lr_col.cpu, av1_dec->lr_col.dma);
>> -    av1_dec->lr_col.cpu = NULL;
>> -}
>> -
>> -static int rockchip_vpu981_av1_dec_tiles_reallocate(struct 
>> hantro_ctx *ctx)
>> -{
>> -    struct hantro_dev *vpu = ctx->dev;
>> -    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> -    struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>> -    const struct v4l2_av1_tile_info *tile_info = 
>> &ctrls->frame->tile_info;
>> -    unsigned int num_tile_cols = tile_info->tile_cols;
>> -    unsigned int height = ALIGN(ctrls->frame->frame_height_minus_1 + 
>> 1, 64);
>> -    unsigned int height_in_sb = height / 64;
>> -    unsigned int stripe_num = ((height + 8) + 63) / 64;
>> -    size_t size;
>> -
>> -    if (av1_dec->db_data_col.size >=
>> -        ALIGN(height * 12 * ctx->bit_depth / 8, 128) * num_tile_cols)
>> -        return 0;
>> -
>> -    rockchip_vpu981_av1_dec_tiles_free(ctx);
>> -
>> -    size = ALIGN(height * 12 * ctx->bit_depth / 8, 128) * 
>> num_tile_cols;
>> -    av1_dec->db_data_col.cpu = dma_alloc_coherent(vpu->dev, size,
>> -                              &av1_dec->db_data_col.dma,
>> -                              GFP_KERNEL);
>> -    if (!av1_dec->db_data_col.cpu)
>> -        goto buffer_allocation_error;
>> -    av1_dec->db_data_col.size = size;
>> -
>> -    size = ALIGN(height * 2 * 16 / 4, 128) * num_tile_cols;
>> -    av1_dec->db_ctrl_col.cpu = dma_alloc_coherent(vpu->dev, size,
>> -                              &av1_dec->db_ctrl_col.dma,
>> -                              GFP_KERNEL);
>> -    if (!av1_dec->db_ctrl_col.cpu)
>> -        goto buffer_allocation_error;
>> -    av1_dec->db_ctrl_col.size = size;
>> -
>> -    size = ALIGN(height_in_sb * 44 * ctx->bit_depth * 16 / 8, 128) * 
>> num_tile_cols;
>> -    av1_dec->cdef_col.cpu = dma_alloc_coherent(vpu->dev, size,
>> -                           &av1_dec->cdef_col.dma,
>> -                           GFP_KERNEL);
>> -    if (!av1_dec->cdef_col.cpu)
>> -        goto buffer_allocation_error;
>> -    av1_dec->cdef_col.size = size;
>> -
>> -    size = ALIGN(height_in_sb * (3040 + 1280), 128) * num_tile_cols;
>> -    av1_dec->sr_col.cpu = dma_alloc_coherent(vpu->dev, size,
>> -                         &av1_dec->sr_col.dma,
>> -                         GFP_KERNEL);
>> -    if (!av1_dec->sr_col.cpu)
>> -        goto buffer_allocation_error;
>> -    av1_dec->sr_col.size = size;
>> -
>> -    size = ALIGN(stripe_num * 1536 * ctx->bit_depth / 8, 128) * 
>> num_tile_cols;
>> -    av1_dec->lr_col.cpu = dma_alloc_coherent(vpu->dev, size,
>> -                         &av1_dec->lr_col.dma,
>> -                         GFP_KERNEL);
>> -    if (!av1_dec->lr_col.cpu)
>> -        goto buffer_allocation_error;
>> -    av1_dec->lr_col.size = size;
>> -
>> -    av1_dec->num_tile_cols_allocated = num_tile_cols;
>> -    return 0;
>> -
>> -buffer_allocation_error:
>> -    rockchip_vpu981_av1_dec_tiles_free(ctx);
>> -    return -ENOMEM;
>> -}
>> -
>> -void rockchip_vpu981_av1_dec_exit(struct hantro_ctx *ctx)
>> -{
>> -    struct hantro_dev *vpu = ctx->dev;
>> -    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> -
>> -    if (av1_dec->global_model.cpu)
>> -        dma_free_coherent(vpu->dev, av1_dec->global_model.size,
>> -                  av1_dec->global_model.cpu,
>> -                  av1_dec->global_model.dma);
>> -    av1_dec->global_model.cpu = NULL;
>> -
>> -    if (av1_dec->tile_info.cpu)
>> -        dma_free_coherent(vpu->dev, av1_dec->tile_info.size,
>> -                  av1_dec->tile_info.cpu,
>> -                  av1_dec->tile_info.dma);
>> -    av1_dec->tile_info.cpu = NULL;
>> -
>> -    if (av1_dec->film_grain.cpu)
>> -        dma_free_coherent(vpu->dev, av1_dec->film_grain.size,
>> -                  av1_dec->film_grain.cpu,
>> -                  av1_dec->film_grain.dma);
>> -    av1_dec->film_grain.cpu = NULL;
>> -
>> -    if (av1_dec->prob_tbl.cpu)
>> -        dma_free_coherent(vpu->dev, av1_dec->prob_tbl.size,
>> -                  av1_dec->prob_tbl.cpu, av1_dec->prob_tbl.dma);
>> -    av1_dec->prob_tbl.cpu = NULL;
>> -
>> -    if (av1_dec->prob_tbl_out.cpu)
>> -        dma_free_coherent(vpu->dev, av1_dec->prob_tbl_out.size,
>> -                  av1_dec->prob_tbl_out.cpu,
>> -                  av1_dec->prob_tbl_out.dma);
>> -    av1_dec->prob_tbl_out.cpu = NULL;
>> -
>> -    if (av1_dec->tile_buf.cpu)
>> -        dma_free_coherent(vpu->dev, av1_dec->tile_buf.size,
>> -                  av1_dec->tile_buf.cpu, av1_dec->tile_buf.dma);
>> -    av1_dec->tile_buf.cpu = NULL;
>> -
>> -    rockchip_vpu981_av1_dec_tiles_free(ctx);
>> -}
>> -
>> -int rockchip_vpu981_av1_dec_init(struct hantro_ctx *ctx)
>> -{
>> -    struct hantro_dev *vpu = ctx->dev;
>> -    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> -
>> -    memset(av1_dec, 0, sizeof(*av1_dec));
>> -
>> -    av1_dec->global_model.cpu = dma_alloc_coherent(vpu->dev, 
>> GLOBAL_MODEL_SIZE,
>> - &av1_dec->global_model.dma,
>> -                               GFP_KERNEL);
>> -    if (!av1_dec->global_model.cpu)
>> -        return -ENOMEM;
>> -    av1_dec->global_model.size = GLOBAL_MODEL_SIZE;
>> -
>> -    av1_dec->tile_info.cpu = dma_alloc_coherent(vpu->dev, 
>> AV1_TILE_INFO_SIZE,
>> -                            &av1_dec->tile_info.dma,
>> -                            GFP_KERNEL);
>> -    if (!av1_dec->tile_info.cpu)
>> -        return -ENOMEM;
>> -    av1_dec->tile_info.size = AV1_TILE_INFO_SIZE;
>> -
>> -    av1_dec->film_grain.cpu = dma_alloc_coherent(vpu->dev,
>> -                             ALIGN(sizeof(struct 
>> rockchip_av1_film_grain), 2048),
>> -                             &av1_dec->film_grain.dma,
>> -                             GFP_KERNEL);
>> -    if (!av1_dec->film_grain.cpu)
>> -        return -ENOMEM;
>> -    av1_dec->film_grain.size = ALIGN(sizeof(struct 
>> rockchip_av1_film_grain), 2048);
>> -
>> -    av1_dec->prob_tbl.cpu = dma_alloc_coherent(vpu->dev,
>> -                           ALIGN(sizeof(struct av1cdfs), 2048),
>> -                           &av1_dec->prob_tbl.dma,
>> -                           GFP_KERNEL);
>> -    if (!av1_dec->prob_tbl.cpu)
>> -        return -ENOMEM;
>> -    av1_dec->prob_tbl.size = ALIGN(sizeof(struct av1cdfs), 2048);
>> -
>> -    av1_dec->prob_tbl_out.cpu = dma_alloc_coherent(vpu->dev,
>> -                               ALIGN(sizeof(struct av1cdfs), 2048),
>> - &av1_dec->prob_tbl_out.dma,
>> -                               GFP_KERNEL);
>> -    if (!av1_dec->prob_tbl_out.cpu)
>> -        return -ENOMEM;
>> -    av1_dec->prob_tbl_out.size = ALIGN(sizeof(struct av1cdfs), 2048);
>> -    av1_dec->cdfs = &av1_dec->default_cdfs;
>> -    av1_dec->cdfs_ndvc = &av1_dec->default_cdfs_ndvc;
>> -
>> -    rockchip_av1_set_default_cdfs(av1_dec->cdfs, av1_dec->cdfs_ndvc);
>> -
>> -    av1_dec->tile_buf.cpu = dma_alloc_coherent(vpu->dev,
>> -                           AV1_TILE_SIZE,
>> -                           &av1_dec->tile_buf.dma,
>> -                           GFP_KERNEL);
>> -    if (!av1_dec->tile_buf.cpu)
>> -        return -ENOMEM;
>> -    av1_dec->tile_buf.size = AV1_TILE_SIZE;
>> -
>> -    return 0;
>> -}
>> -
>> -static int rockchip_vpu981_av1_dec_prepare_run(struct hantro_ctx *ctx)
>> -{
>> -    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> -    struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>> -
>> -    ctrls->sequence = hantro_get_ctrl(ctx, 
>> V4L2_CID_STATELESS_AV1_SEQUENCE);
>> -    if (WARN_ON(!ctrls->sequence))
>> -        return -EINVAL;
>> -
>> -    ctrls->tile_group_entry =
>> -        hantro_get_ctrl(ctx, V4L2_CID_STATELESS_AV1_TILE_GROUP_ENTRY);
>> -    if (WARN_ON(!ctrls->tile_group_entry))
>> -        return -EINVAL;
>> -
>> -    ctrls->frame = hantro_get_ctrl(ctx, V4L2_CID_STATELESS_AV1_FRAME);
>> -    if (WARN_ON(!ctrls->frame))
>> -        return -EINVAL;
>> -
>> -    ctrls->film_grain =
>> -        hantro_get_ctrl(ctx, V4L2_CID_STATELESS_AV1_FILM_GRAIN);
>> -
>> -    return rockchip_vpu981_av1_dec_tiles_reallocate(ctx);
>> -}
>> -
>> -static inline int rockchip_vpu981_av1_dec_get_msb(u32 n)
>> -{
>> -    if (n == 0)
>> -        return 0;
>> -    return 31 ^ __builtin_clz(n);
>> -}
>> -
>> -static short rockchip_vpu981_av1_dec_resolve_divisor_32(u32 d, short 
>> *shift)
>> -{
>> -    int f;
>> -    u64 e;
>> -
>> -    *shift = rockchip_vpu981_av1_dec_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
>> -rockchip_vpu981_av1_dec_get_shear_params(const u32 *params, s64 *alpha,
>> -                     s64 *beta, s64 *gamma, s64 *delta)
>> -{
>> -    const int *mat = params;
>> -    short shift;
>> -    short y;
>> -    long long gv, dv;
>> -
>> -    if (mat[2] <= 0)
>> -        return;
>> -
>> -    *alpha = clamp_val(mat[2] - (1 << WARPEDMODEL_PREC_BITS), 
>> S16_MIN, S16_MAX);
>> -    *beta = clamp_val(mat[3], S16_MIN, S16_MAX);
>> -
>> -    y = rockchip_vpu981_av1_dec_resolve_divisor_32(abs(mat[2]), 
>> &shift) * (mat[2] < 0 ? -1 : 1);
>> -
>> -    gv = ((long long)mat[4] * (1 << WARPEDMODEL_PREC_BITS)) * y;
>> -
>> -    *gamma = clamp_val((int)AV1_DIV_ROUND_UP_POW2_SIGNED(gv, shift), 
>> S16_MIN, S16_MAX);
>> -
>> -    dv = ((long long)mat[3] * mat[4]) * y;
>> -    *delta = clamp_val(mat[5] -
>> -        (int)AV1_DIV_ROUND_UP_POW2_SIGNED(dv, shift) - (1 << 
>> WARPEDMODEL_PREC_BITS),
>> -        S16_MIN, S16_MAX);
>> -
>> -    *alpha = AV1_DIV_ROUND_UP_POW2_SIGNED(*alpha, 
>> WARP_PARAM_REDUCE_BITS)
>> -         * (1 << WARP_PARAM_REDUCE_BITS);
>> -    *beta = AV1_DIV_ROUND_UP_POW2_SIGNED(*beta, WARP_PARAM_REDUCE_BITS)
>> -        * (1 << WARP_PARAM_REDUCE_BITS);
>> -    *gamma = AV1_DIV_ROUND_UP_POW2_SIGNED(*gamma, 
>> WARP_PARAM_REDUCE_BITS)
>> -         * (1 << WARP_PARAM_REDUCE_BITS);
>> -    *delta = AV1_DIV_ROUND_UP_POW2_SIGNED(*delta, 
>> WARP_PARAM_REDUCE_BITS)
>> -        * (1 << WARP_PARAM_REDUCE_BITS);
>> -}
>>     static void rockchip_vpu981_av1_dec_set_global_model(struct 
>> hantro_ctx *ctx)
>>   {
>>       struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> -    struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>> -    const struct v4l2_ctrl_av1_frame *frame = ctrls->frame;
>> -    const struct v4l2_av1_global_motion *gm = &frame->global_motion;
>> -    u8 *dst = av1_dec->global_model.cpu;
>>       struct hantro_dev *vpu = ctx->dev;
>> -    int ref_frame, i;
>> -
>> -    memset(dst, 0, GLOBAL_MODEL_SIZE);
>> -    for (ref_frame = 0; ref_frame < V4L2_AV1_REFS_PER_FRAME; 
>> ++ref_frame) {
>> -        s64 alpha = 0, beta = 0, gamma = 0, delta = 0;
>> -
>> -        for (i = 0; i < 6; ++i) {
>> -            if (i == 2)
>> -                *(s32 *)dst =
>> -                    gm->params[V4L2_AV1_REF_LAST_FRAME + ref_frame][3];
>> -            else if (i == 3)
>> -                *(s32 *)dst =
>> -                    gm->params[V4L2_AV1_REF_LAST_FRAME + ref_frame][2];
>> -            else
>> -                *(s32 *)dst =
>> -                    gm->params[V4L2_AV1_REF_LAST_FRAME + ref_frame][i];
>> -            dst += 4;
>> -        }
>> -
>> -        if (gm->type[V4L2_AV1_REF_LAST_FRAME + ref_frame] <= 
>> V4L2_AV1_WARP_MODEL_AFFINE)
>> - 
>> rockchip_vpu981_av1_dec_get_shear_params(&gm->params[V4L2_AV1_REF_LAST_FRAME 
>> + ref_frame][0],
>> -                                 &alpha, &beta, &gamma, &delta);
>> -
>> -        *(s16 *)dst = alpha;
>> -        dst += 2;
>> -        *(s16 *)dst = beta;
>> -        dst += 2;
>> -        *(s16 *)dst = gamma;
>> -        dst += 2;
>> -        *(s16 *)dst = delta;
>> -        dst += 2;
>> -    }
>>   +    hantro_av1_set_global_model(ctx);
>>       hantro_write_addr(vpu, AV1_GLOBAL_MODEL, 
>> av1_dec->global_model.dma);
>>   }
>>   -static int rockchip_vpu981_av1_tile_log2(int target)
>> -{
>> -    int k;
>> -
>> -    /*
>> -     * returns the smallest value for k such that 1 << k is greater
>> -     * than or equal to target
>> -     */
>> -    for (k = 0; (1 << k) < target; k++);
>> -
>> -    return k;
>> -}
>> -
>>   static void rockchip_vpu981_av1_dec_set_tile_info(struct hantro_ctx 
>> *ctx)
>>   {
>>       struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>>       struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>>       const struct v4l2_av1_tile_info *tile_info = 
>> &ctrls->frame->tile_info;
>> -    const struct v4l2_ctrl_av1_tile_group_entry *group_entry =
>> -        ctrls->tile_group_entry;
>>       int context_update_y =
>>           tile_info->context_update_tile_id / tile_info->tile_cols;
>>       int context_update_x =
>>           tile_info->context_update_tile_id % tile_info->tile_cols;
>>       int context_update_tile_id =
>>           context_update_x * tile_info->tile_rows + context_update_y;
>> -    u8 *dst = av1_dec->tile_info.cpu;
>>       struct hantro_dev *vpu = ctx->dev;
>> -    int tile0, tile1;
>> -
>> -    memset(dst, 0, av1_dec->tile_info.size);
>> -
>> -    for (tile0 = 0; tile0 < tile_info->tile_cols; tile0++) {
>> -        for (tile1 = 0; tile1 < tile_info->tile_rows; tile1++) {
>> -            int tile_id = tile1 * tile_info->tile_cols + tile0;
>> -            u32 start, end;
>> -            u32 y0 =
>> -                tile_info->height_in_sbs_minus_1[tile1] + 1;
>> -            u32 x0 = tile_info->width_in_sbs_minus_1[tile0] + 1;
>> -
>> -            /* tile size in SB units (width,height) */
>> -            *dst++ = x0;
>> -            *dst++ = 0;
>> -            *dst++ = 0;
>> -            *dst++ = 0;
>> -            *dst++ = y0;
>> -            *dst++ = 0;
>> -            *dst++ = 0;
>> -            *dst++ = 0;
>> -
>> -            /* tile start position */
>> -            start = group_entry[tile_id].tile_offset - 
>> group_entry[0].tile_offset;
>> -            *dst++ = start & 255;
>> -            *dst++ = (start >> 8) & 255;
>> -            *dst++ = (start >> 16) & 255;
>> -            *dst++ = (start >> 24) & 255;
>> -
>> -            /* number of bytes in tile data */
>> -            end = start + group_entry[tile_id].tile_size;
>> -            *dst++ = end & 255;
>> -            *dst++ = (end >> 8) & 255;
>> -            *dst++ = (end >> 16) & 255;
>> -            *dst++ = (end >> 24) & 255;
>> -        }
>> -    }
>> +
>> +    hantro_av1_set_tile_info(ctx);
>>         hantro_reg_write(vpu, &av1_multicore_expect_context_update, 
>> !!(context_update_x == 0));
>>       hantro_reg_write(vpu, &av1_tile_enable,
>> @@ -631,8 +44,8 @@ static void 
>> rockchip_vpu981_av1_dec_set_tile_info(struct hantro_ctx *ctx)
>>       hantro_reg_write(vpu, &av1_num_tile_rows_8k, 
>> tile_info->tile_rows);
>>       hantro_reg_write(vpu, &av1_context_update_tile_id, 
>> context_update_tile_id);
>>       hantro_reg_write(vpu, &av1_tile_transpose, 1);
>> -    if (rockchip_vpu981_av1_tile_log2(tile_info->tile_cols) ||
>> -        rockchip_vpu981_av1_tile_log2(tile_info->tile_rows))
>> +    if (hantro_av1_tile_log2(tile_info->tile_cols) ||
>> +        hantro_av1_tile_log2(tile_info->tile_rows))
>>           hantro_reg_write(vpu, &av1_dec_tile_size_mag, 
>> tile_info->tile_size_bytes - 1);
>>       else
>>           hantro_reg_write(vpu, &av1_dec_tile_size_mag, 3);
>> @@ -640,50 +53,6 @@ static void 
>> rockchip_vpu981_av1_dec_set_tile_info(struct hantro_ctx *ctx)
>>       hantro_write_addr(vpu, AV1_TILE_BASE, av1_dec->tile_info.dma);
>>   }
>>   -static int rockchip_vpu981_av1_dec_get_dist(struct hantro_ctx *ctx,
>> -                        int a, int b)
>> -{
>> -    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> -    struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>> -    int bits = ctrls->sequence->order_hint_bits - 1;
>> -    int diff, m;
>> -
>> -    if (!ctrls->sequence->order_hint_bits)
>> -        return 0;
>> -
>> -    diff = a - b;
>> -    m = 1 << bits;
>> -    diff = (diff & (m - 1)) - (diff & m);
>> -
>> -    return diff;
>> -}
>> -
>> -static void rockchip_vpu981_av1_dec_set_frame_sign_bias(struct 
>> hantro_ctx *ctx)
>> -{
>> -    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> -    struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>> -    const struct v4l2_ctrl_av1_frame *frame = ctrls->frame;
>> -    const struct v4l2_ctrl_av1_sequence *sequence = ctrls->sequence;
>> -    int i;
>> -
>> -    if (!sequence->order_hint_bits || IS_INTRA(frame->frame_type)) {
>> -        for (i = 0; i < V4L2_AV1_TOTAL_REFS_PER_FRAME; i++)
>> -            av1_dec->ref_frame_sign_bias[i] = 0;
>> -
>> -        return;
>> -    }
>> -    // Identify the nearest forward and backward references.
>> -    for (i = 0; i < V4L2_AV1_TOTAL_REFS_PER_FRAME - 1; i++) {
>> -        if (rockchip_vpu981_get_frame_index(ctx, i) >= 0) {
>> -            int rel_off =
>> -                rockchip_vpu981_av1_dec_get_dist(ctx,
>> - rockchip_vpu981_get_order_hint(ctx, i),
>> -                                 frame->order_hint);
>> -            av1_dec->ref_frame_sign_bias[i + 1] = (rel_off <= 0) ? 0 
>> : 1;
>> -        }
>> -    }
>> -}
>> -
>>   static bool
>>   rockchip_vpu981_av1_dec_set_ref(struct hantro_ctx *ctx, int ref, 
>> int idx,
>>                   int width, int height)
>> @@ -806,12 +175,12 @@ static void 
>> rockchip_vpu981_av1_dec_set_segmentation(struct hantro_ctx *ctx)
>>         if (!!(seg->flags & V4L2_AV1_SEGMENTATION_FLAG_ENABLED) &&
>>           frame->primary_ref_frame < V4L2_AV1_REFS_PER_FRAME) {
>> -        int idx = rockchip_vpu981_get_frame_index(ctx, 
>> frame->primary_ref_frame);
>> +        int idx = hantro_av1_get_frame_index(ctx, 
>> frame->primary_ref_frame);
>>   -        if (idx >= 0) {
>> +        if (idx != AV1_INVALID_IDX) {
>>               dma_addr_t luma_addr, mv_addr = 0;
>>               struct hantro_decoded_buffer *seg;
>> -            size_t mv_offset = 
>> rockchip_vpu981_av1_dec_chroma_size(ctx);
>> +            size_t mv_offset = hantro_av1_chroma_size(ctx);
>>                 seg = 
>> vb2_to_hantro_decoded_buf(&av1_dec->frame_refs[idx].vb2_ref->vb2_buf);
>>               luma_addr = hantro_get_dec_buf_addr(ctx, 
>> &seg->base.vb.vb2_buf);
>> @@ -1041,35 +410,6 @@ static void 
>> rockchip_vpu981_av1_dec_set_segmentation(struct hantro_ctx *ctx)
>>                segval[7][V4L2_AV1_SEG_LVL_REF_GLOBALMV]);
>>   }
>>   -static bool rockchip_vpu981_av1_dec_is_lossless(struct hantro_ctx 
>> *ctx)
>> -{
>> -    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> -    struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>> -    const struct v4l2_ctrl_av1_frame *frame = ctrls->frame;
>> -    const struct v4l2_av1_segmentation *segmentation = 
>> &frame->segmentation;
>> -    const struct v4l2_av1_quantization *quantization = 
>> &frame->quantization;
>> -    int i;
>> -
>> -    for (i = 0; i < V4L2_AV1_MAX_SEGMENTS; i++) {
>> -        int qindex = quantization->base_q_idx;
>> -
>> -        if (segmentation->feature_enabled[i] &
>> - V4L2_AV1_SEGMENT_FEATURE_ENABLED(V4L2_AV1_SEG_LVL_ALT_Q)) {
>> -            qindex += 
>> segmentation->feature_data[i][V4L2_AV1_SEG_LVL_ALT_Q];
>> -        }
>> -        qindex = clamp(qindex, 0, 255);
>> -
>> -        if (qindex ||
>> -            quantization->delta_q_y_dc ||
>> -            quantization->delta_q_u_dc ||
>> -            quantization->delta_q_u_ac ||
>> -            quantization->delta_q_v_dc ||
>> -            quantization->delta_q_v_ac)
>> -            return false;
>> -    }
>> -    return true;
>> -}
>> -
>>   static void rockchip_vpu981_av1_dec_set_loopfilter(struct 
>> hantro_ctx *ctx)
>>   {
>>       struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> @@ -1089,7 +429,7 @@ static void 
>> rockchip_vpu981_av1_dec_set_loopfilter(struct hantro_ctx *ctx)
>>       hantro_reg_write(vpu, &av1_filt_level3, loop_filter->level[3]);
>>         if (loop_filter->flags & 
>> V4L2_AV1_LOOP_FILTER_FLAG_DELTA_ENABLED &&
>> -        !rockchip_vpu981_av1_dec_is_lossless(ctx) &&
>> +        !hantro_av1_is_lossless(ctx) &&
>>           !(frame->flags & V4L2_AV1_FRAME_FLAG_ALLOW_INTRABC)) {
>>           hantro_reg_write(vpu, &av1_filt_ref_adj_0,
>>                    loop_filter->ref_deltas[0]);
>> @@ -1128,121 +468,27 @@ static void 
>> rockchip_vpu981_av1_dec_set_loopfilter(struct hantro_ctx *ctx)
>>       hantro_write_addr(vpu, AV1_DB_CTRL_COL, av1_dec->db_ctrl_col.dma);
>>   }
>>   -static void rockchip_vpu981_av1_dec_update_prob(struct hantro_ctx 
>> *ctx)
>> -{
>> -    struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> -    struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>> -    const struct v4l2_ctrl_av1_frame *frame = ctrls->frame;
>> -    bool frame_is_intra = IS_INTRA(frame->frame_type);
>> -    struct av1cdfs *out_cdfs = (struct av1cdfs 
>> *)av1_dec->prob_tbl_out.cpu;
>> -    int i;
>> -
>> -    if (frame->flags & 
>> V4L2_AV1_FRAME_FLAG_DISABLE_FRAME_END_UPDATE_CDF)
>> -        return;
>> -
>> -    for (i = 0; i < NUM_REF_FRAMES; i++) {
>> -        if (frame->refresh_frame_flags & BIT(i)) {
>> -            struct mvcdfs stored_mv_cdf;
>> -
>> -            rockchip_av1_get_cdfs(ctx, i);
>> -            stored_mv_cdf = av1_dec->cdfs->mv_cdf;
>> -            *av1_dec->cdfs = *out_cdfs;
>> -            if (frame_is_intra) {
>> -                av1_dec->cdfs->mv_cdf = stored_mv_cdf;
>> -                *av1_dec->cdfs_ndvc = out_cdfs->mv_cdf;
>> -            }
>> -            rockchip_av1_store_cdfs(ctx,
>> -                        frame->refresh_frame_flags);
>> -            break;
>> -        }
>> -    }
>> -}
>> -
>> -void rockchip_vpu981_av1_dec_done(struct hantro_ctx *ctx)
>> -{
>> -    rockchip_vpu981_av1_dec_update_prob(ctx);
>> -}
>> -
>>   static void rockchip_vpu981_av1_dec_set_prob(struct hantro_ctx *ctx)
>>   {
>>       struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>> -    struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>> -    const struct v4l2_ctrl_av1_frame *frame = ctrls->frame;
>> -    const struct v4l2_av1_quantization *quantization = 
>> &frame->quantization;
>>       struct hantro_dev *vpu = ctx->dev;
>> -    bool error_resilient_mode =
>> -        !!(frame->flags & V4L2_AV1_FRAME_FLAG_ERROR_RESILIENT_MODE);
>> -    bool frame_is_intra = IS_INTRA(frame->frame_type);
>> -
>> -    if (error_resilient_mode || frame_is_intra ||
>> -        frame->primary_ref_frame == AV1_PRIMARY_REF_NONE) {
>> -        av1_dec->cdfs = &av1_dec->default_cdfs;
>> -        av1_dec->cdfs_ndvc = &av1_dec->default_cdfs_ndvc;
>> - rockchip_av1_default_coeff_probs(quantization->base_q_idx,
>> -                         av1_dec->cdfs);
>> -    } else {
>> -        rockchip_av1_get_cdfs(ctx, 
>> frame->ref_frame_idx[frame->primary_ref_frame]);
>> -    }
>> -    rockchip_av1_store_cdfs(ctx, frame->refresh_frame_flags);
>> -
>> -    memcpy(av1_dec->prob_tbl.cpu, av1_dec->cdfs, sizeof(struct 
>> av1cdfs));
>> -
>> -    if (frame_is_intra) {
>> -        int mv_offset = offsetof(struct av1cdfs, mv_cdf);
>> -        /* Overwrite MV context area with intrabc MV context */
>> -        memcpy(av1_dec->prob_tbl.cpu + mv_offset, av1_dec->cdfs_ndvc,
>> -               sizeof(struct mvcdfs));
>> -    }
>>   +    hantro_av1_set_prob(ctx);
>>       hantro_write_addr(vpu, AV1_PROP_TABLE_OUT, 
>> av1_dec->prob_tbl_out.dma);
>>       hantro_write_addr(vpu, AV1_PROP_TABLE, av1_dec->prob_tbl.dma);
>>   }
>>   -static void
>> -rockchip_vpu981_av1_dec_init_scaling_function(const u8 *values, 
>> const u8 *scaling,
>> -                          u8 num_points, u8 *scaling_lut)
>> -{
>> -    int i, point;
>> -
>> -    if (num_points == 0) {
>> -        memset(scaling_lut, 0, 256);
>> -        return;
>> -    }
>> -
>> -    for (point = 0; point < num_points - 1; point++) {
>> -        int x;
>> -        s32 delta_y = scaling[point + 1] - scaling[point];
>> -        s32 delta_x = values[point + 1] - values[point];
>> -        s64 delta =
>> -            delta_x ? delta_y * ((65536 + (delta_x >> 1)) /
>> -                     delta_x) : 0;
>> -
>> -        for (x = 0; x < delta_x; x++) {
>> -            scaling_lut[values[point] + x] =
>> -                scaling[point] +
>> -                (s32)((x * delta + 32768) >> 16);
>> -        }
>> -    }
>> -
>> -    for (i = values[num_points - 1]; i < 256; i++)
>> -        scaling_lut[i] = scaling[num_points - 1];
>> -}
>>     static void rockchip_vpu981_av1_dec_set_fgs(struct hantro_ctx *ctx)
>>   {
>>       struct hantro_av1_dec_hw_ctx *av1_dec = &ctx->av1_dec;
>>       struct hantro_av1_dec_ctrls *ctrls = &av1_dec->ctrls;
>>       const struct v4l2_ctrl_av1_film_grain *film_grain = 
>> ctrls->film_grain;
>> -    struct rockchip_av1_film_grain *fgmem = av1_dec->film_grain.cpu;
>> +    struct hantro_av1_film_grain *fgmem = av1_dec->film_grain.cpu;
>>       struct hantro_dev *vpu = ctx->dev;
>>       bool scaling_from_luma =
>>           !!(film_grain->flags & 
>> V4L2_AV1_FILM_GRAIN_FLAG_CHROMA_SCALING_FROM_LUMA);
>> -    s32 (*ar_coeffs_y)[24];
>> -    s32 (*ar_coeffs_cb)[25];
>> -    s32 (*ar_coeffs_cr)[25];
>> -    s32 (*luma_grain_block)[73][82];
>> -    s32 (*cb_grain_block)[38][44];
>> -    s32 (*cr_grain_block)[38][44];
>> +    struct hantro_av1_coeffs_grain_block *coeffs;
>>       s32 ar_coeff_lag, ar_coeff_shift;
>>       s32 grain_scale_shift, bitdepth;
>>       s32 grain_center, grain_min, grain_max;
>> @@ -1269,18 +515,9 @@ static void 
>> rockchip_vpu981_av1_dec_set_fgs(struct hantro_ctx *ctx)
>>           return;
>>       }
>>   -    ar_coeffs_y = kzalloc(sizeof(int32_t) * 24, GFP_KERNEL);
>> -    ar_coeffs_cb = kzalloc(sizeof(int32_t) * 25, GFP_KERNEL);
>> -    ar_coeffs_cr = kzalloc(sizeof(int32_t) * 25, GFP_KERNEL);
>> -    luma_grain_block = kzalloc(sizeof(int32_t) * 73 * 82, GFP_KERNEL);
>> -    cb_grain_block = kzalloc(sizeof(int32_t) * 38 * 44, GFP_KERNEL);
>> -    cr_grain_block = kzalloc(sizeof(int32_t) * 38 * 44, GFP_KERNEL);
>> -
>> -    if (!ar_coeffs_y || !ar_coeffs_cb || !ar_coeffs_cr ||
>> -        !luma_grain_block || !cb_grain_block || !cr_grain_block) {
>> -        pr_warn("Fail allocating memory for film grain parameters\n");
>> -        goto alloc_fail;
>> -    }
>> +    coeffs = kzalloc(sizeof(*coeffs), GFP_KERNEL);
>> +    if (!coeffs)
>> +        return;
>>         hantro_reg_write(vpu, &av1_apply_grain, 1);
>>   @@ -1316,10 +553,10 @@ static void 
>> rockchip_vpu981_av1_dec_set_fgs(struct hantro_ctx *ctx)
>>       hantro_reg_write(vpu, &av1_chroma_scaling_from_luma, 
>> scaling_from_luma);
>>       hantro_reg_write(vpu, &av1_random_seed, film_grain->grain_seed);
>>   - 
>> rockchip_vpu981_av1_dec_init_scaling_function(film_grain->point_y_value,
>> -                              film_grain->point_y_scaling,
>> -                              film_grain->num_y_points,
>> -                              fgmem->scaling_lut_y);
>> + hantro_av1_init_scaling_function(film_grain->point_y_value,
>> +                     film_grain->point_y_scaling,
>> +                     film_grain->num_y_points,
>> +                     fgmem->scaling_lut_y);
>>         if (film_grain->flags &
>>           V4L2_AV1_FILM_GRAIN_FLAG_CHROMA_SCALING_FROM_LUMA) {
>> @@ -1328,19 +565,19 @@ static void 
>> rockchip_vpu981_av1_dec_set_fgs(struct hantro_ctx *ctx)
>>           memcpy(fgmem->scaling_lut_cr, fgmem->scaling_lut_y,
>>                  sizeof(*fgmem->scaling_lut_y) * 256);
>>       } else {
>> -        rockchip_vpu981_av1_dec_init_scaling_function
>> +        hantro_av1_init_scaling_function
>>               (film_grain->point_cb_value, film_grain->point_cb_scaling,
>>                film_grain->num_cb_points, fgmem->scaling_lut_cb);
>> -        rockchip_vpu981_av1_dec_init_scaling_function
>> +        hantro_av1_init_scaling_function
>>               (film_grain->point_cr_value, film_grain->point_cr_scaling,
>>                film_grain->num_cr_points, fgmem->scaling_lut_cr);
>>       }
>>         for (i = 0; i < V4L2_AV1_AR_COEFFS_SIZE; i++) {
>>           if (i < 24)
>> -            (*ar_coeffs_y)[i] = film_grain->ar_coeffs_y_plus_128[i] 
>> - 128;
>> -        (*ar_coeffs_cb)[i] = film_grain->ar_coeffs_cb_plus_128[i] - 
>> 128;
>> -        (*ar_coeffs_cr)[i] = film_grain->ar_coeffs_cr_plus_128[i] - 
>> 128;
>> +            coeffs->ar_coeffs_y[i] = 
>> film_grain->ar_coeffs_y_plus_128[i] - 128;
>> +        coeffs->ar_coeffs_cb[i] = 
>> film_grain->ar_coeffs_cb_plus_128[i] - 128;
>> +        coeffs->ar_coeffs_cr[i] = 
>> film_grain->ar_coeffs_cr_plus_128[i] - 128;
>>       }
>>         ar_coeff_lag = film_grain->ar_coeff_lag;
>> @@ -1351,46 +588,38 @@ static void 
>> rockchip_vpu981_av1_dec_set_fgs(struct hantro_ctx *ctx)
>>       grain_min = 0 - grain_center;
>>       grain_max = (256 << (bitdepth - 8)) - 1 - grain_center;
>>   -    rockchip_av1_generate_luma_grain_block(luma_grain_block, 
>> bitdepth,
>> -                           film_grain->num_y_points, grain_scale_shift,
>> -                           ar_coeff_lag, ar_coeffs_y, ar_coeff_shift,
>> -                           grain_min, grain_max, 
>> film_grain->grain_seed);
>> -
>> -    rockchip_av1_generate_chroma_grain_block(luma_grain_block, 
>> cb_grain_block,
>> -                         cr_grain_block, bitdepth,
>> -                         film_grain->num_y_points,
>> -                         film_grain->num_cb_points,
>> -                         film_grain->num_cr_points,
>> -                         grain_scale_shift, ar_coeff_lag, ar_coeffs_cb,
>> -                         ar_coeffs_cr, ar_coeff_shift, grain_min,
>> -                         grain_max,
>> -                         scaling_from_luma,
>> -                         film_grain->grain_seed);
>> +    hantro_av1_generate_luma_grain_block(coeffs, bitdepth,
>> +                         film_grain->num_y_points, grain_scale_shift,
>> +                         ar_coeff_lag, ar_coeff_shift,
>> +                         grain_min, grain_max, film_grain->grain_seed);
>> +
>> +    hantro_av1_generate_chroma_grain_block(coeffs, bitdepth,
>> +                           film_grain->num_y_points,
>> +                           film_grain->num_cb_points,
>> +                           film_grain->num_cr_points,
>> +                           grain_scale_shift, ar_coeff_lag,
>> +                           ar_coeff_shift, grain_min,
>> +                           grain_max,
>> +                           scaling_from_luma,
>> +                           film_grain->grain_seed);
>>         for (i = 0; i < 64; i++) {
>>           for (j = 0; j < 64; j++)
>>               fgmem->cropped_luma_grain_block[i * 64 + j] =
>> -                (*luma_grain_block)[i + 9][j + 9];
>> +                coeffs->luma_grain_block[i + 9][j + 9];
>>       }
>>         for (i = 0; i < 32; i++) {
>>           for (j = 0; j < 32; j++) {
>>               fgmem->cropped_chroma_grain_block[i * 64 + 2 * j] =
>> -                (*cb_grain_block)[i + 6][j + 6];
>> +                coeffs->cb_grain_block[i + 6][j + 6];
>>               fgmem->cropped_chroma_grain_block[i * 64 + 2 * j + 1] =
>> -                (*cr_grain_block)[i + 6][j + 6];
>> +                coeffs->cr_grain_block[i + 6][j + 6];
>>           }
>>       }
>>   +    kfree(coeffs);
>>       hantro_write_addr(vpu, AV1_FILM_GRAIN, av1_dec->film_grain.dma);
>> -
>> -alloc_fail:
>> -    kfree(ar_coeffs_y);
>> -    kfree(ar_coeffs_cb);
>> -    kfree(ar_coeffs_cr);
>> -    kfree(luma_grain_block);
>> -    kfree(cb_grain_block);
>> -    kfree(cr_grain_block);
>>   }
>>     static void rockchip_vpu981_av1_dec_set_cdef(struct hantro_ctx *ctx)
>> @@ -1617,13 +846,13 @@ static void 
>> rockchip_vpu981_av1_dec_set_other_frames(struct hantro_ctx *ctx)
>>       int ref_ind = 0;
>>       int rf, idx;
>>   -    alt_frame_offset = rockchip_vpu981_get_order_hint(ctx, 
>> ALT_BUF_IDX);
>> -    gld_frame_offset = rockchip_vpu981_get_order_hint(ctx, 
>> GLD_BUF_IDX);
>> -    bwd_frame_offset = rockchip_vpu981_get_order_hint(ctx, 
>> BWD_BUF_IDX);
>> -    alt2_frame_offset = rockchip_vpu981_get_order_hint(ctx, 
>> ALT2_BUF_IDX);
>> +    alt_frame_offset = hantro_av1_get_order_hint(ctx, ALT_BUF_IDX);
>> +    gld_frame_offset = hantro_av1_get_order_hint(ctx, GLD_BUF_IDX);
>> +    bwd_frame_offset = hantro_av1_get_order_hint(ctx, BWD_BUF_IDX);
>> +    alt2_frame_offset = hantro_av1_get_order_hint(ctx, ALT2_BUF_IDX);
>>   -    idx = rockchip_vpu981_get_frame_index(ctx, LST_BUF_IDX);
>> -    if (idx >= 0) {
>> +    idx = hantro_av1_get_frame_index(ctx, LST_BUF_IDX);
>> +    if (idx != AV1_INVALID_IDX) {
>>           int alt_frame_offset_in_lst =
>> av1_dec->frame_refs[idx].order_hints[V4L2_AV1_REF_ALTREF_FRAME];
>>           bool is_lst_overlay =
>> @@ -1644,8 +873,9 @@ static void 
>> rockchip_vpu981_av1_dec_set_other_frames(struct hantro_ctx *ctx)
>>           ref_stamp--;
>>       }
>>   -    idx = rockchip_vpu981_get_frame_index(ctx, BWD_BUF_IDX);
>> -    if (rockchip_vpu981_av1_dec_get_dist(ctx, bwd_frame_offset, 
>> cur_frame_offset) > 0) {
>> +    idx = hantro_av1_get_frame_index(ctx, BWD_BUF_IDX);
>> +    if (idx != AV1_INVALID_IDX &&
>> +        hantro_av1_get_dist(ctx, bwd_frame_offset, cur_frame_offset) 
>> > 0) {
>>           int bwd_mi_cols = av1_dec->frame_refs[idx].mi_cols;
>>           int bwd_mi_rows = av1_dec->frame_refs[idx].mi_rows;
>>           bool bwd_intra_only =
>> @@ -1659,8 +889,9 @@ static void 
>> rockchip_vpu981_av1_dec_set_other_frames(struct hantro_ctx *ctx)
>>           }
>>       }
>>   -    idx = rockchip_vpu981_get_frame_index(ctx, ALT2_BUF_IDX);
>> -    if (rockchip_vpu981_av1_dec_get_dist(ctx, alt2_frame_offset, 
>> cur_frame_offset) > 0) {
>> +    idx = hantro_av1_get_frame_index(ctx, ALT2_BUF_IDX);
>> +    if (idx != AV1_INVALID_IDX &&
>> +        hantro_av1_get_dist(ctx, alt2_frame_offset, 
>> cur_frame_offset) > 0) {
>>           int alt2_mi_cols = av1_dec->frame_refs[idx].mi_cols;
>>           int alt2_mi_rows = av1_dec->frame_refs[idx].mi_rows;
>>           bool alt2_intra_only =
>> @@ -1674,8 +905,9 @@ static void 
>> rockchip_vpu981_av1_dec_set_other_frames(struct hantro_ctx *ctx)
>>           }
>>       }
>>   -    idx = rockchip_vpu981_get_frame_index(ctx, ALT_BUF_IDX);
>> -    if (rockchip_vpu981_av1_dec_get_dist(ctx, alt_frame_offset, 
>> cur_frame_offset) > 0 &&
>> +    idx = hantro_av1_get_frame_index(ctx, ALT_BUF_IDX);
>> +    if (idx != AV1_INVALID_IDX &&
>> +        hantro_av1_get_dist(ctx, alt_frame_offset, cur_frame_offset) 
>> > 0 &&
>>           ref_stamp >= 0) {
>>           int alt_mi_cols = av1_dec->frame_refs[idx].mi_cols;
>>           int alt_mi_rows = av1_dec->frame_refs[idx].mi_rows;
>> @@ -1690,8 +922,8 @@ static void 
>> rockchip_vpu981_av1_dec_set_other_frames(struct hantro_ctx *ctx)
>>           }
>>       }
>>   -    idx = rockchip_vpu981_get_frame_index(ctx, LST2_BUF_IDX);
>> -    if (idx >= 0 && ref_stamp >= 0) {
>> +    idx = hantro_av1_get_frame_index(ctx, LST2_BUF_IDX);
>> +    if (idx != AV1_INVALID_IDX && ref_stamp >= 0) {
>>           int lst2_mi_cols = av1_dec->frame_refs[idx].mi_cols;
>>           int lst2_mi_rows = av1_dec->frame_refs[idx].mi_rows;
>>           bool lst2_intra_only =
>> @@ -1706,14 +938,14 @@ static void 
>> rockchip_vpu981_av1_dec_set_other_frames(struct hantro_ctx *ctx)
>>       }
>>         for (rf = 0; rf < V4L2_AV1_TOTAL_REFS_PER_FRAME - 1; ++rf) {
>> -        idx = rockchip_vpu981_get_frame_index(ctx, rf);
>> -        if (idx >= 0) {
>> -            int rf_order_hint = rockchip_vpu981_get_order_hint(ctx, 
>> rf);
>> +        idx = hantro_av1_get_frame_index(ctx, rf);
>> +        if (idx != AV1_INVALID_IDX) {
>> +            int rf_order_hint = hantro_av1_get_order_hint(ctx, rf);
>>                 cur_offset[rf] =
>> -                rockchip_vpu981_av1_dec_get_dist(ctx, 
>> cur_frame_offset, rf_order_hint);
>> +                hantro_av1_get_dist(ctx, cur_frame_offset, 
>> rf_order_hint);
>>               cur_roffset[rf] =
>> -                rockchip_vpu981_av1_dec_get_dist(ctx, rf_order_hint, 
>> cur_frame_offset);
>> +                hantro_av1_get_dist(ctx, rf_order_hint, 
>> cur_frame_offset);
>>           } else {
>>               cur_offset[rf] = 0;
>>               cur_roffset[rf] = 0;
>> @@ -1736,35 +968,41 @@ static void 
>> rockchip_vpu981_av1_dec_set_other_frames(struct hantro_ctx *ctx)
>>       if (use_ref_frame_mvs && ref_ind > 0 &&
>>           cur_offset[mf_types[0] - V4L2_AV1_REF_LAST_FRAME] <= 
>> MAX_FRAME_DISTANCE &&
>>           cur_offset[mf_types[0] - V4L2_AV1_REF_LAST_FRAME] >= 
>> -MAX_FRAME_DISTANCE) {
>> -        int rf = rockchip_vpu981_get_order_hint(ctx, refs_selected[0]);
>> -        int idx = rockchip_vpu981_get_frame_index(ctx, 
>> refs_selected[0]);
>> -        u32 *oh = av1_dec->frame_refs[idx].order_hints;
>> +        int rf = hantro_av1_get_order_hint(ctx, refs_selected[0]);
>> +        int idx = hantro_av1_get_frame_index(ctx, refs_selected[0]);
>> +        u32 *oh;
>>           int val;
>>   +        if (idx == AV1_INVALID_IDX)
>> +            goto second_ref;
>> +
>> +        oh = av1_dec->frame_refs[idx].order_hints;
>> +
>>           hantro_reg_write(vpu, &av1_use_temporal0_mvs, 1);
>>   -        val = rockchip_vpu981_av1_dec_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_LAST_FRAME]);
>> +        val = hantro_av1_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_LAST_FRAME]);
>>           hantro_reg_write(vpu, &av1_mf1_last_offset, val);
>>   -        val = rockchip_vpu981_av1_dec_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_LAST2_FRAME]);
>> +        val = hantro_av1_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_LAST2_FRAME]);
>>           hantro_reg_write(vpu, &av1_mf1_last2_offset, val);
>>   -        val = rockchip_vpu981_av1_dec_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_LAST3_FRAME]);
>> +        val = hantro_av1_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_LAST3_FRAME]);
>>           hantro_reg_write(vpu, &av1_mf1_last3_offset, val);
>>   -        val = rockchip_vpu981_av1_dec_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_GOLDEN_FRAME]);
>> +        val = hantro_av1_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_GOLDEN_FRAME]);
>>           hantro_reg_write(vpu, &av1_mf1_golden_offset, val);
>>   -        val = rockchip_vpu981_av1_dec_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_BWDREF_FRAME]);
>> +        val = hantro_av1_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_BWDREF_FRAME]);
>>           hantro_reg_write(vpu, &av1_mf1_bwdref_offset, val);
>>   -        val = rockchip_vpu981_av1_dec_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_ALTREF2_FRAME]);
>> +        val = hantro_av1_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_ALTREF2_FRAME]);
>>           hantro_reg_write(vpu, &av1_mf1_altref2_offset, val);
>>   -        val = rockchip_vpu981_av1_dec_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_ALTREF_FRAME]);
>> +        val = hantro_av1_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_ALTREF_FRAME]);
>>           hantro_reg_write(vpu, &av1_mf1_altref_offset, val);
>>       }
>>   +second_ref:
>>       hantro_reg_write(vpu, &av1_mf2_last_offset, 0);
>>       hantro_reg_write(vpu, &av1_mf2_last2_offset, 0);
>>       hantro_reg_write(vpu, &av1_mf2_last3_offset, 0);
>> @@ -1776,35 +1014,41 @@ static void 
>> rockchip_vpu981_av1_dec_set_other_frames(struct hantro_ctx *ctx)
>>       if (use_ref_frame_mvs && ref_ind > 1 &&
>>           cur_offset[mf_types[1] - V4L2_AV1_REF_LAST_FRAME] <= 
>> MAX_FRAME_DISTANCE &&
>>           cur_offset[mf_types[1] - V4L2_AV1_REF_LAST_FRAME] >= 
>> -MAX_FRAME_DISTANCE) {
>> -        int rf = rockchip_vpu981_get_order_hint(ctx, refs_selected[1]);
>> -        int idx = rockchip_vpu981_get_frame_index(ctx, 
>> refs_selected[1]);
>> -        u32 *oh = av1_dec->frame_refs[idx].order_hints;
>> +        int rf = hantro_av1_get_order_hint(ctx, refs_selected[1]);
>> +        int idx = hantro_av1_get_frame_index(ctx, refs_selected[1]);
>> +        u32 *oh;
>>           int val;
>>   +        if (idx == AV1_INVALID_IDX)
>> +            goto third_ref;
>> +
>> +        oh = av1_dec->frame_refs[idx].order_hints;
>> +
>>           hantro_reg_write(vpu, &av1_use_temporal1_mvs, 1);
>>   -        val = rockchip_vpu981_av1_dec_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_LAST_FRAME]);
>> +        val = hantro_av1_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_LAST_FRAME]);
>>           hantro_reg_write(vpu, &av1_mf2_last_offset, val);
>>   -        val = rockchip_vpu981_av1_dec_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_LAST2_FRAME]);
>> +        val = hantro_av1_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_LAST2_FRAME]);
>>           hantro_reg_write(vpu, &av1_mf2_last2_offset, val);
>>   -        val = rockchip_vpu981_av1_dec_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_LAST3_FRAME]);
>> +        val = hantro_av1_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_LAST3_FRAME]);
>>           hantro_reg_write(vpu, &av1_mf2_last3_offset, val);
>>   -        val = rockchip_vpu981_av1_dec_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_GOLDEN_FRAME]);
>> +        val = hantro_av1_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_GOLDEN_FRAME]);
>>           hantro_reg_write(vpu, &av1_mf2_golden_offset, val);
>>   -        val = rockchip_vpu981_av1_dec_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_BWDREF_FRAME]);
>> +        val = hantro_av1_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_BWDREF_FRAME]);
>>           hantro_reg_write(vpu, &av1_mf2_bwdref_offset, val);
>>   -        val = rockchip_vpu981_av1_dec_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_ALTREF2_FRAME]);
>> +        val = hantro_av1_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_ALTREF2_FRAME]);
>>           hantro_reg_write(vpu, &av1_mf2_altref2_offset, val);
>>   -        val = rockchip_vpu981_av1_dec_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_ALTREF_FRAME]);
>> +        val = hantro_av1_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_ALTREF_FRAME]);
>>           hantro_reg_write(vpu, &av1_mf2_altref_offset, val);
>>       }
>>   +third_ref:
>>       hantro_reg_write(vpu, &av1_mf3_last_offset, 0);
>>       hantro_reg_write(vpu, &av1_mf3_last2_offset, 0);
>>       hantro_reg_write(vpu, &av1_mf3_last3_offset, 0);
>> @@ -1816,35 +1060,41 @@ static void 
>> rockchip_vpu981_av1_dec_set_other_frames(struct hantro_ctx *ctx)
>>       if (use_ref_frame_mvs && ref_ind > 2 &&
>>           cur_offset[mf_types[2] - V4L2_AV1_REF_LAST_FRAME] <= 
>> MAX_FRAME_DISTANCE &&
>>           cur_offset[mf_types[2] - V4L2_AV1_REF_LAST_FRAME] >= 
>> -MAX_FRAME_DISTANCE) {
>> -        int rf = rockchip_vpu981_get_order_hint(ctx, refs_selected[2]);
>> -        int idx = rockchip_vpu981_get_frame_index(ctx, 
>> refs_selected[2]);
>> -        u32 *oh = av1_dec->frame_refs[idx].order_hints;
>> +        int rf = hantro_av1_get_order_hint(ctx, refs_selected[2]);
>> +        int idx = hantro_av1_get_frame_index(ctx, refs_selected[2]);
>> +        u32 *oh;
>>           int val;
>>   +        if (idx == AV1_INVALID_IDX)
>> +            goto offsets;
>> +
>> +        oh = av1_dec->frame_refs[idx].order_hints;
>> +
>>           hantro_reg_write(vpu, &av1_use_temporal2_mvs, 1);
>>   -        val = rockchip_vpu981_av1_dec_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_LAST_FRAME]);
>> +        val = hantro_av1_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_LAST_FRAME]);
>>           hantro_reg_write(vpu, &av1_mf3_last_offset, val);
>>   -        val = rockchip_vpu981_av1_dec_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_LAST2_FRAME]);
>> +        val = hantro_av1_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_LAST2_FRAME]);
>>           hantro_reg_write(vpu, &av1_mf3_last2_offset, val);
>>   -        val = rockchip_vpu981_av1_dec_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_LAST3_FRAME]);
>> +        val = hantro_av1_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_LAST3_FRAME]);
>>           hantro_reg_write(vpu, &av1_mf3_last3_offset, val);
>>   -        val = rockchip_vpu981_av1_dec_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_GOLDEN_FRAME]);
>> +        val = hantro_av1_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_GOLDEN_FRAME]);
>>           hantro_reg_write(vpu, &av1_mf3_golden_offset, val);
>>   -        val = rockchip_vpu981_av1_dec_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_BWDREF_FRAME]);
>> +        val = hantro_av1_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_BWDREF_FRAME]);
>>           hantro_reg_write(vpu, &av1_mf3_bwdref_offset, val);
>>   -        val = rockchip_vpu981_av1_dec_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_ALTREF2_FRAME]);
>> +        val = hantro_av1_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_ALTREF2_FRAME]);
>>           hantro_reg_write(vpu, &av1_mf3_altref2_offset, val);
>>   -        val = rockchip_vpu981_av1_dec_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_ALTREF_FRAME]);
>> +        val = hantro_av1_get_dist(ctx, rf, 
>> oh[V4L2_AV1_REF_ALTREF_FRAME]);
>>           hantro_reg_write(vpu, &av1_mf3_altref_offset, val);
>>       }
>>   +offsets:
>>       hantro_reg_write(vpu, &av1_cur_last_offset, cur_offset[0]);
>>       hantro_reg_write(vpu, &av1_cur_last2_offset, cur_offset[1]);
>>       hantro_reg_write(vpu, &av1_cur_last3_offset, cur_offset[2]);
>> @@ -1883,9 +1133,9 @@ static void 
>> rockchip_vpu981_av1_dec_set_reference_frames(struct hantro_ctx *ctx)
>>         if (!allow_intrabc) {
>>           for (i = 0; i < V4L2_AV1_REFS_PER_FRAME; i++) {
>> -            int idx = rockchip_vpu981_get_frame_index(ctx, i);
>> +            int idx = hantro_av1_get_frame_index(ctx, i);
>>   -            if (idx >= 0)
>> +            if (idx != AV1_INVALID_IDX)
>>                   ref_count[idx]++;
>>           }
>>   @@ -1898,7 +1148,7 @@ static void 
>> rockchip_vpu981_av1_dec_set_reference_frames(struct hantro_ctx *ctx)
>>       }
>>       hantro_reg_write(vpu, &av1_ref_frames, ref_frames);
>>   -    rockchip_vpu981_av1_dec_set_frame_sign_bias(ctx);
>> +    hantro_av1_set_frame_sign_bias(ctx);
>>         for (i = V4L2_AV1_REF_LAST_FRAME; i < 
>> V4L2_AV1_TOTAL_REFS_PER_FRAME; i++) {
>>           u32 ref = i - 1;
>> @@ -1910,8 +1160,8 @@ static void 
>> rockchip_vpu981_av1_dec_set_reference_frames(struct hantro_ctx *ctx)
>>               width = frame->frame_width_minus_1 + 1;
>>               height = frame->frame_height_minus_1 + 1;
>>           } else {
>> -            if (rockchip_vpu981_get_frame_index(ctx, ref) > 0)
>> -                idx = rockchip_vpu981_get_frame_index(ctx, ref);
>> +            if (hantro_av1_get_frame_index(ctx, ref) != 
>> AV1_INVALID_IDX)
>> +                idx = hantro_av1_get_frame_index(ctx, ref);
>>               width = av1_dec->frame_refs[idx].width;
>>               height = av1_dec->frame_refs[idx].height;
>>           }
>> @@ -1943,20 +1193,6 @@ static void 
>> rockchip_vpu981_av1_dec_set_reference_frames(struct hantro_ctx *ctx)
>>       rockchip_vpu981_av1_dec_set_other_frames(ctx);
>>   }
>>   -static int rockchip_vpu981_av1_get_hardware_tx_mode(enum 
>> v4l2_av1_tx_mode tx_mode)
>> -{
>> -    switch (tx_mode) {
>> -    case V4L2_AV1_TX_MODE_ONLY_4X4:
>> -        return ROCKCHIP_AV1_TX_MODE_ONLY_4X4;
>> -    case V4L2_AV1_TX_MODE_LARGEST:
>> -        return ROCKCHIP_AV1_TX_MODE_32x32;
>> -    case V4L2_AV1_TX_MODE_SELECT:
>> -        return ROCKCHIP_AV1_TX_MODE_SELECT;
>> -    }
>> -
>> -    return ROCKCHIP_AV1_TX_MODE_32x32;
>> -}
>> -
>>   static void rockchip_vpu981_av1_dec_set_parameters(struct 
>> hantro_ctx *ctx)
>>   {
>>       struct hantro_dev *vpu = ctx->dev;
>> @@ -2029,7 +1265,7 @@ static void 
>> rockchip_vpu981_av1_dec_set_parameters(struct hantro_ctx *ctx)
>>       hantro_reg_write(vpu, &av1_comp_pred_mode,
>>                (ctrls->frame->flags & 
>> V4L2_AV1_FRAME_FLAG_REFERENCE_SELECT) ? 2 : 0);
>>   -    tx_mode = 
>> rockchip_vpu981_av1_get_hardware_tx_mode(ctrls->frame->tx_mode);
>> +    tx_mode = hantro_av1_get_hardware_tx_mode(ctrls->frame->tx_mode);
>>       hantro_reg_write(vpu, &av1_transform_mode, tx_mode);
>>       hantro_reg_write(vpu, &av1_max_cb_size,
>>                (ctrls->sequence->flags
>> @@ -2061,7 +1297,7 @@ static void 
>> rockchip_vpu981_av1_dec_set_parameters(struct hantro_ctx *ctx)
>>           hantro_reg_write(vpu, &av1_qmlevel_v, 0xff);
>>       }
>>   -    hantro_reg_write(vpu, &av1_lossless_e, 
>> rockchip_vpu981_av1_dec_is_lossless(ctx));
>> +    hantro_reg_write(vpu, &av1_lossless_e, 
>> hantro_av1_is_lossless(ctx));
>>       hantro_reg_write(vpu, &av1_quant_delta_v_dc, 
>> ctrls->frame->quantization.delta_q_v_dc);
>>       hantro_reg_write(vpu, &av1_quant_delta_v_ac, 
>> ctrls->frame->quantization.delta_q_v_ac);
>>   @@ -2109,8 +1345,8 @@ 
>> rockchip_vpu981_av1_dec_set_output_buffer(struct hantro_ctx *ctx)
>>       struct hantro_decoded_buffer *dst;
>>       struct vb2_v4l2_buffer *vb2_dst;
>>       dma_addr_t luma_addr, chroma_addr, mv_addr = 0;
>> -    size_t cr_offset = rockchip_vpu981_av1_dec_luma_size(ctx);
>> -    size_t mv_offset = rockchip_vpu981_av1_dec_chroma_size(ctx);
>> +    size_t cr_offset = hantro_av1_luma_size(ctx);
>> +    size_t mv_offset = hantro_av1_chroma_size(ctx);
>>         vb2_dst = 
>> av1_dec->frame_refs[av1_dec->current_frame_index].vb2_ref;
>>       dst = vb2_to_hantro_decoded_buf(&vb2_dst->vb2_buf);
>> @@ -2134,7 +1370,7 @@ int rockchip_vpu981_av1_dec_run(struct 
>> hantro_ctx *ctx)
>>         hantro_start_prepare_run(ctx);
>>   -    ret = rockchip_vpu981_av1_dec_prepare_run(ctx);
>> +    ret = hantro_av1_prepare_run(ctx);
>>       if (ret)
>>           goto prepare_error;
>>   @@ -2144,8 +1380,8 @@ int rockchip_vpu981_av1_dec_run(struct 
>> hantro_ctx *ctx)
>>           goto prepare_error;
>>       }
>>   -    rockchip_vpu981_av1_dec_clean_refs(ctx);
>> -    rockchip_vpu981_av1_dec_frame_ref(ctx, vb2_src->vb2_buf.timestamp);
>> +    hantro_av1_clean_refs(ctx);
>> +    hantro_av1_frame_ref(ctx, vb2_src->vb2_buf.timestamp);
>>         rockchip_vpu981_av1_dec_set_parameters(ctx);
>>       rockchip_vpu981_av1_dec_set_global_model(ctx);
>> diff --git a/drivers/media/platform/verisilicon/rockchip_vpu_hw.c 
>> b/drivers/media/platform/verisilicon/rockchip_vpu_hw.c
>> index 02673be9878e..f50a3e38097e 100644
>> --- a/drivers/media/platform/verisilicon/rockchip_vpu_hw.c
>> +++ b/drivers/media/platform/verisilicon/rockchip_vpu_hw.c
>> @@ -9,6 +9,7 @@
>>   #include <linux/clk.h>
>>     #include "hantro.h"
>> +#include "hantro_av1.h"
>>   #include "hantro_jpeg.h"
>>   #include "hantro_g1_regs.h"
>>   #include "hantro_h1_regs.h"
>> @@ -608,9 +609,9 @@ static const struct hantro_codec_ops 
>> rk3568_vepu_codec_ops[] = {
>>   static const struct hantro_codec_ops rk3588_vpu981_codec_ops[] = {
>>       [HANTRO_MODE_AV1_DEC] = {
>>           .run = rockchip_vpu981_av1_dec_run,
>> -        .init = rockchip_vpu981_av1_dec_init,
>> -        .exit = rockchip_vpu981_av1_dec_exit,
>> -        .done = rockchip_vpu981_av1_dec_done,
>> +        .init = hantro_av1_init,
>> +        .exit = hantro_av1_exit,
>> +        .done = hantro_av1_update_prob,
>>       },
>>   };
>>   /*

_______________________________________________
Linux-rockchip mailing list
Linux-rockchip@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-rockchip

      reply	other threads:[~2026-08-31 13:38 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-05-12  7:59 [PATCH v5] media: verisilicon: Create AV1 helper library Benjamin Gaignard
2026-07-10  9:45 ` Benjamin Gaignard
2026-08-31 13:38   ` Benjamin Gaignard [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=25e17f59-e637-48a2-8a9b-4913738b7c4e@collabora.com \
    --to=benjamin.gaignard@collabora.com \
    --cc=heiko@sntech.de \
    --cc=kernel@collabora.com \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-media@vger.kernel.org \
    --cc=linux-rockchip@lists.infradead.org \
    --cc=mchehab@kernel.org \
    --cc=nicolas.dufresne@collabora.com \
    --cc=p.zabel@pengutronix.de \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox