lkml.org 
[lkml]   [2020]   [Jun]   [2]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
SubjectRe: [PATCH] media: rkvdec: Fix H264 scaling list order
From
Date
Hi Jonas,

Thanks a lot for the fix! Indeed it fixes
a few bitstream that had artifacts :)

On Fri, 2020-05-22 at 20:21 +0000, Jonas Karlman wrote:
> The Rockchip Video Decoder driver is expecting that the values in a
> scaling list are in zig-zag order and applies the inverse scanning process
> to get the values in matrix order.
>
> Commit 0b0393d59eb4 ("media: uapi: h264: clarify expected
> scaling_list_4x4/8x8 order") clarified that the values in the scaling list
> should already be in matrix order.
>
> Fix this by removing the reordering and change to use two memcpy.
>
> Fixes: cd33c830448b ("media: rkvdec: Add the rkvdec driver")
> Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
> ---
> drivers/staging/media/rkvdec/rkvdec-h264.c | 70 +++++++---------------
> 1 file changed, 22 insertions(+), 48 deletions(-)
>
> diff --git a/drivers/staging/media/rkvdec/rkvdec-h264.c b/drivers/staging/media/rkvdec/rkvdec-h264.c
> index cd4980d06be7..2719f0c66a4a 100644
> --- a/drivers/staging/media/rkvdec/rkvdec-h264.c
> +++ b/drivers/staging/media/rkvdec/rkvdec-h264.c
> @@ -18,11 +18,16 @@
> /* Size with u32 units. */
> #define RKV_CABAC_INIT_BUFFER_SIZE (3680 + 128)
> #define RKV_RPS_SIZE ((128 + 128) / 4)
> -#define RKV_SCALING_LIST_SIZE (6 * 16 + 6 * 64 + 128)
> #define RKV_ERROR_INFO_SIZE (256 * 144 * 4)
>
> #define RKVDEC_NUM_REFLIST 3
>
> +struct rkvdec_scaling_matrix {

How about we call this rkvdec_scaling_list
or even rkvdec_h264_scaling_list?

It'll make code more readable and easier to grep.

> + u8 scaling_list_4x4[6][16];
> + u8 scaling_list_8x8[6][64];
> + u8 padding[128];

Oops, something is wrong here, maybe some
mistake when porting code around.

Rockchip MPP defines the scaling list size as:

#define RKV_SCALING_LIST_SIZE (6*16 + 2*64 + 128)

Consistently with the fact that 4:4:4 sampling
is not supported (chroma_format_idc == 3),
only the first two scaling matrices are used.

Moreover, given all these buffers are specified separately
(see below), I bet not even the 128 padding bytes
are needed.

> +};
> +
> struct rkvdec_sps_pps_packet {
> u32 info[8];
> };
> @@ -86,7 +91,7 @@ struct rkvdec_ps_field {
> /* Data structure describing auxiliary buffer format. */
> struct rkvdec_h264_priv_tbl {
> s8 cabac_table[4][464][2];

As can be seen, all these buffers can be allocated
independently, which perhaps could reduce the
pressure on the DMA allocator.

> - u8 scaling_list[RKV_SCALING_LIST_SIZE];
> + struct rkvdec_scaling_matrix scaling_list;
> u32 rps[RKV_RPS_SIZE];
> struct rkvdec_sps_pps_packet param_set[256];
> u8 err_info[RKV_ERROR_INFO_SIZE];

In particular, I wonder if the error info stuff
is really required.

I guess we'll want to merge a simple fix,
so no need to address any of these questions
for now.

With the struct rkvdec_scaling_matrix rename:

Reviewed-by: Ezequiel Garcia <ezequiel@collabora.com>

Thanks,
Ezequiel
> @@ -785,56 +790,25 @@ static void assemble_hw_rps(struct rkvdec_ctx *ctx,
> }
> }
>
> -/*
> - * NOTE: The values in a scaling list are in zig-zag order, apply inverse
> - * scanning process to get the values in matrix order.
> - */
> -static const u32 zig_zag_4x4[16] = {
> - 0, 1, 4, 8, 5, 2, 3, 6, 9, 12, 13, 10, 7, 11, 14, 15
> -};
> -
> -static const u32 zig_zag_8x8[64] = {
> - 0, 1, 8, 16, 9, 2, 3, 10, 17, 24, 32, 25, 18, 11, 4, 5,
> - 12, 19, 26, 33, 40, 48, 41, 34, 27, 20, 13, 6, 7, 14, 21, 28,
> - 35, 42, 49, 56, 57, 50, 43, 36, 29, 22, 15, 23, 30, 37, 44, 51,
> - 58, 59, 52, 45, 38, 31, 39, 46, 53, 60, 61, 54, 47, 55, 62, 63
> -};
> -
> -static void reorder_scaling_list(struct rkvdec_ctx *ctx,
> - struct rkvdec_h264_run *run)
> +static void assemble_hw_scaling_list(struct rkvdec_ctx *ctx,
> + struct rkvdec_h264_run *run)
> {
> const struct v4l2_ctrl_h264_scaling_matrix *scaling = run->scaling_matrix;
> - const size_t num_list_4x4 = ARRAY_SIZE(scaling->scaling_list_4x4);
> - const size_t list_len_4x4 = ARRAY_SIZE(scaling->scaling_list_4x4[0]);
> - const size_t num_list_8x8 = ARRAY_SIZE(scaling->scaling_list_8x8);
> - const size_t list_len_8x8 = ARRAY_SIZE(scaling->scaling_list_8x8[0]);
> struct rkvdec_h264_ctx *h264_ctx = ctx->priv;
> struct rkvdec_h264_priv_tbl *tbl = h264_ctx->priv_tbl.cpu;
> - u8 *dst = tbl->scaling_list;
> - const u8 *src;
> - int i, j;
> -
> - BUILD_BUG_ON(ARRAY_SIZE(zig_zag_4x4) != list_len_4x4);
> - BUILD_BUG_ON(ARRAY_SIZE(zig_zag_8x8) != list_len_8x8);
> - BUILD_BUG_ON(ARRAY_SIZE(tbl->scaling_list) <
> - num_list_4x4 * list_len_4x4 +
> - num_list_8x8 * list_len_8x8);
> -
> - src = &scaling->scaling_list_4x4[0][0];
> - for (i = 0; i < num_list_4x4; ++i) {
> - for (j = 0; j < list_len_4x4; ++j)
> - dst[zig_zag_4x4[j]] = src[j];
> - src += list_len_4x4;
> - dst += list_len_4x4;
> - }
>
> - src = &scaling->scaling_list_8x8[0][0];
> - for (i = 0; i < num_list_8x8; ++i) {
> - for (j = 0; j < list_len_8x8; ++j)
> - dst[zig_zag_8x8[j]] = src[j];
> - src += list_len_8x8;
> - dst += list_len_8x8;
> - }
> + BUILD_BUG_ON(sizeof(tbl->scaling_list.scaling_list_4x4) !=
> + sizeof(scaling->scaling_list_4x4));
> + BUILD_BUG_ON(sizeof(tbl->scaling_list.scaling_list_8x8) !=
> + sizeof(scaling->scaling_list_8x8));
> +
> + memcpy(tbl->scaling_list.scaling_list_4x4,
> + scaling->scaling_list_4x4,
> + sizeof(scaling->scaling_list_4x4));
> +
> + memcpy(tbl->scaling_list.scaling_list_8x8,
> + scaling->scaling_list_8x8,
> + sizeof(scaling->scaling_list_8x8));
> }
>
> /*
> @@ -1126,7 +1100,7 @@ static int rkvdec_h264_run(struct rkvdec_ctx *ctx)
> v4l2_h264_build_b_ref_lists(&reflist_builder, h264_ctx->reflists.b0,
> h264_ctx->reflists.b1);
>
> - reorder_scaling_list(ctx, &run);
> + assemble_hw_scaling_list(ctx, &run);
> assemble_hw_pps(ctx, &run);
> assemble_hw_rps(ctx, &run);
> config_registers(ctx, &run);


\
 
 \ /
  Last update: 2020-06-02 06:37    [W:0.159 / U:0.044 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site