[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <9bb1910c-2a6e-49e6-e839-7b677974acf3@collabora.com>
Date: Mon, 7 Jun 2021 15:27:41 +0200
From: Benjamin Gaignard <benjamin.gaignard@...labora.com>
To: Ezequiel Garcia <ezequiel@...labora.com>, hverkuil@...all.nl,
p.zabel@...gutronix.de, mchehab@...nel.org, shawnguo@...nel.org,
s.hauer@...gutronix.de, festevam@...il.com,
gregkh@...uxfoundation.org, mripard@...nel.org,
paul.kocialkowski@...tlin.com, wens@...e.org,
jernej.skrabec@...l.net, emil.l.velikov@...il.com,
andrzej.p@...labora.com, jc@...esim.co.uk
Cc: kernel@...gutronix.de, linux-imx@....com,
linux-media@...r.kernel.org, linux-rockchip@...ts.infradead.org,
linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 2/8] media: hantro: Add support of compressed reference
buffers
Le 04/06/2021 à 18:32, Ezequiel Garcia a écrit :
> Hi Benjamin,
>
> Thanks for the patches, they looks really nice!
>
> On Fri, 2021-06-04 at 15:06 +0200, Benjamin Gaignard wrote:
>> Reference frame compression is a feature added in G2 decoder to compress
>> frame buffers so that the bandwidth of storing/loading reference frames
>> can be reduced, especially when the resolution of decoded stream is of
>> high definition.
>>
> Can you do some tracing using the performance counters in patch 1,
> with and without this patch, while we are here, it would be interesting
> to run some experiments.
>
> If I understand correctly, the G2 HEVC engine is producing tiled buffers,
> which are then post-processed to NV12.
>
> So this compression feature is compressing the tiled buffers,
> the NV12 are still uncompressed. If this is the case, could add
> some comments somewhere about the different sets of buffers involved?
I will do that
>
> Also, I believe I mentioned this before, is there a plan
> to be able to expose the tiled buffers to userspace (i.e. without
> the NV12 linear post-processing)? I think the display controller
> can take compressed tiled-buffers.
Yes it could be possible to use vb2 buffers instead of internally
allocated buffers but that would means define a new format/modifier
before doing that.
>
>> Signed-off-by: Benjamin Gaignard <benjamin.gaignard@...labora.com>
>> ---
>> .../staging/media/hantro/hantro_g2_hevc_dec.c | 20 ++++++++++--
>> drivers/staging/media/hantro/hantro_g2_regs.h | 4 +++
>> drivers/staging/media/hantro/hantro_hevc.c | 32 ++++++++++++++++++-
>> drivers/staging/media/hantro/hantro_hw.h | 2 ++
>> 4 files changed, 54 insertions(+), 4 deletions(-)
>>
>> diff --git a/drivers/staging/media/hantro/hantro_g2_hevc_dec.c b/drivers/staging/media/hantro/hantro_g2_hevc_dec.c
>> index 89fac5146433..9a715e803037 100644
>> --- a/drivers/staging/media/hantro/hantro_g2_hevc_dec.c
>> +++ b/drivers/staging/media/hantro/hantro_g2_hevc_dec.c
>> @@ -382,10 +382,12 @@ static int set_ref(struct hantro_ctx *ctx)
>> const struct v4l2_ctrl_hevc_pps *pps = ctrls->pps;
>> const struct v4l2_ctrl_hevc_decode_params *decode_params = ctrls->decode_params;
>> const struct v4l2_hevc_dpb_entry *dpb = decode_params->dpb;
>> - dma_addr_t luma_addr, chroma_addr, mv_addr = 0;
>> + dma_addr_t luma_addr, chroma_addr, mv_addr, compress_luma_addr, compress_chroma_addr = 0;
>> struct hantro_dev *vpu = ctx->dev;
>> size_t cr_offset = hantro_hevc_chroma_offset(sps);
>> size_t mv_offset = hantro_hevc_motion_vectors_offset(sps);
>> + size_t compress_luma_offset = hantro_hevc_luma_compress_offset(sps);
>> + size_t compress_chroma_offset = hantro_hevc_chroma_compress_offset(sps);
>> u32 max_ref_frames;
>> u16 dpb_longterm_e;
>> static const struct hantro_reg cur_poc[] = {
>> @@ -460,6 +462,8 @@ static int set_ref(struct hantro_ctx *ctx)
>>
>> chroma_addr = luma_addr + cr_offset;
>> mv_addr = luma_addr + mv_offset;
>> + compress_luma_addr = luma_addr + compress_luma_offset;
>> + compress_chroma_addr = luma_addr + compress_chroma_offset;
>>
>> if (dpb[i].rps == V4L2_HEVC_DPB_ENTRY_RPS_LT_CURR)
>> dpb_longterm_e |= BIT(V4L2_HEVC_DPB_ENTRIES_NUM_MAX - 1 - i);
>> @@ -467,6 +471,8 @@ static int set_ref(struct hantro_ctx *ctx)
>> hantro_write_addr(vpu, G2_REG_ADDR_REF(i), luma_addr);
>> hantro_write_addr(vpu, G2_REG_CHR_REF(i), chroma_addr);
>> hantro_write_addr(vpu, G2_REG_DMV_REF(i), mv_addr);
>> + hantro_write_addr(vpu, G2_COMP_ADDR_REF(i), compress_luma_addr);
>> + hantro_write_addr(vpu, G2_COMP_CHR_REF(i), compress_chroma_addr);
> I'm a bit confused here, you are seeing luma_addr and also compress_luma_addr
> for references??
Yes both are need to make this work. It also done like that in vendor stack.
>
> I'm not saying this is wrong, just would like some comments in the code
> so the intention is a bit more clear and I can understand what's going on.
I will add comments to clarify that point.
Benjamin
>
>> }
>>
>> luma_addr = hantro_hevc_get_ref_buf(ctx, decode_params->pic_order_cnt_val);
>> @@ -475,7 +481,11 @@ static int set_ref(struct hantro_ctx *ctx)
>>
>> chroma_addr = luma_addr + cr_offset;
>> mv_addr = luma_addr + mv_offset;
>> + compress_luma_addr = luma_addr + compress_luma_offset;
>> + compress_chroma_addr = luma_addr + compress_chroma_offset;
>>
>> + hantro_write_addr(vpu, G2_COMP_ADDR_REF(i), compress_luma_addr);
>> + hantro_write_addr(vpu, G2_COMP_CHR_REF(i), compress_chroma_addr);
>> hantro_write_addr(vpu, G2_REG_ADDR_REF(i), luma_addr);
>> hantro_write_addr(vpu, G2_REG_CHR_REF(i), chroma_addr);
>> hantro_write_addr(vpu, G2_REG_DMV_REF(i++), mv_addr);
>> @@ -483,6 +493,8 @@ static int set_ref(struct hantro_ctx *ctx)
>> hantro_write_addr(vpu, G2_ADDR_DST, luma_addr);
>> hantro_write_addr(vpu, G2_ADDR_DST_CHR, chroma_addr);
>> hantro_write_addr(vpu, G2_ADDR_DST_MV, mv_addr);
>> + hantro_write_addr(vpu, G2_COMP_ADDR_DST, compress_luma_addr);
>> + hantro_write_addr(vpu, G2_COMP_CHR, compress_chroma_addr);
>>
>> hantro_hevc_ref_remove_unused(ctx);
>>
>> @@ -490,6 +502,8 @@ static int set_ref(struct hantro_ctx *ctx)
>> hantro_write_addr(vpu, G2_REG_ADDR_REF(i), 0);
>> hantro_write_addr(vpu, G2_REG_CHR_REF(i), 0);
>> hantro_write_addr(vpu, G2_REG_DMV_REF(i), 0);
>> + hantro_write_addr(vpu, G2_COMP_ADDR_REF(i), 0);
>> + hantro_write_addr(vpu, G2_COMP_CHR_REF(i), 0);
>> }
>>
>> hantro_reg_write(vpu, &g2_refer_lterm_e, dpb_longterm_e);
>> @@ -580,8 +594,8 @@ int hantro_g2_hevc_dec_run(struct hantro_ctx *ctx)
>> /* Don't disable output */
>> hantro_reg_write(vpu, &g2_out_dis, 0);
>>
>> - /* Don't compress buffers */
>> - hantro_reg_write(vpu, &g2_ref_compress_bypass, 1);
>> + /* Compress buffers */
>> + hantro_reg_write(vpu, &g2_ref_compress_bypass, 0);
>>
>> /* use NV12 as output format */
>> hantro_reg_write(vpu, &g2_out_rs_e, 1);
>> diff --git a/drivers/staging/media/hantro/hantro_g2_regs.h b/drivers/staging/media/hantro/hantro_g2_regs.h
>> index 17d84ec9c5c2..0414d92e3860 100644
>> --- a/drivers/staging/media/hantro/hantro_g2_regs.h
>> +++ b/drivers/staging/media/hantro/hantro_g2_regs.h
>> @@ -192,6 +192,10 @@
>> #define G2_TILE_FILTER (G2_SWREG(179))
>> #define G2_TILE_SAO (G2_SWREG(181))
>> #define G2_TILE_BSD (G2_SWREG(183))
>> +#define G2_COMP_ADDR_DST (G2_SWREG(190))
>> +#define G2_COMP_ADDR_REF(i) (G2_SWREG(192) + ((i) * 0x8))
>> +#define G2_COMP_CHR (G2_SWREG(224))
>> +#define G2_COMP_CHR_REF(i) (G2_SWREG(226) + ((i) * 0x8))
>>
>> #define g2_strm_buffer_len G2_DEC_REG(258, 0, 0xffffffff)
>> #define g2_strm_start_offset G2_DEC_REG(259, 0, 0xffffffff)
>> diff --git a/drivers/staging/media/hantro/hantro_hevc.c b/drivers/staging/media/hantro/hantro_hevc.c
>> index 5347f5a41c2a..1b2da990fbf0 100644
>> --- a/drivers/staging/media/hantro/hantro_hevc.c
>> +++ b/drivers/staging/media/hantro/hantro_hevc.c
>> @@ -61,12 +61,42 @@ static size_t hantro_hevc_mv_size(const struct v4l2_ctrl_hevc_sps *sps)
>> return mv_size;
>> }
>>
>> +size_t hantro_hevc_luma_compress_offset(const struct v4l2_ctrl_hevc_sps *sps)
>> +{
>> + return hantro_hevc_motion_vectors_offset(sps) + hantro_hevc_mv_size(sps);
>> +}
>> +
>> +static size_t hantro_hevc_luma_compress_size(const struct v4l2_ctrl_hevc_sps *sps)
>> +{
>> + u32 pic_width_in_cbsy =
>> + round_up((sps->pic_width_in_luma_samples + 8 - 1) / 8, 16);
>> + u32 pic_height_in_cbsy = (sps->pic_height_in_luma_samples + 8 - 1) / 8;
>> +
>> + return round_up(pic_width_in_cbsy * pic_height_in_cbsy, 16);
>> +}
>> +
>> +size_t hantro_hevc_chroma_compress_offset(const struct v4l2_ctrl_hevc_sps *sps)
>> +{
>> + return hantro_hevc_luma_compress_offset(sps) + hantro_hevc_luma_compress_size(sps);
>> +}
>> +
>> +static size_t hantro_hevc_chroma_compress_size(const struct v4l2_ctrl_hevc_sps *sps)
>> +{
>> + u32 pic_width_in_cbsc =
>> + round_up((sps->pic_width_in_luma_samples + 16 - 1) / 16, 16);
>> + u32 pic_height_in_cbsc = (sps->pic_height_in_luma_samples / 2 + 4 - 1) / 4;
>> +
> Can we get some #defines for these magic numbers?
>
>> + return round_up(pic_width_in_cbsc * pic_height_in_cbsc, 16);
>> +}
>> +
>> static size_t hantro_hevc_ref_size(struct hantro_ctx *ctx)
>> {
>> const struct hantro_hevc_dec_ctrls *ctrls = &ctx->hevc_dec.ctrls;
>> const struct v4l2_ctrl_hevc_sps *sps = ctrls->sps;
>>
>> - return hantro_hevc_motion_vectors_offset(sps) + hantro_hevc_mv_size(sps);
>> + return hantro_hevc_motion_vectors_offset(sps) + hantro_hevc_mv_size(sps) +
>> + hantro_hevc_luma_compress_size(sps) +
>> + hantro_hevc_chroma_compress_size(sps);
>> }
>>
>> static void hantro_hevc_ref_free(struct hantro_ctx *ctx)
>> diff --git a/drivers/staging/media/hantro/hantro_hw.h b/drivers/staging/media/hantro/hantro_hw.h
>> index 8fa0aacb61cd..c5374cd74d66 100644
>> --- a/drivers/staging/media/hantro/hantro_hw.h
>> +++ b/drivers/staging/media/hantro/hantro_hw.h
>> @@ -246,6 +246,8 @@ dma_addr_t hantro_hevc_get_ref_buf(struct hantro_ctx *ctx, int poc);
>> void hantro_hevc_ref_remove_unused(struct hantro_ctx *ctx);
>> size_t hantro_hevc_chroma_offset(const struct v4l2_ctrl_hevc_sps *sps);
>> size_t hantro_hevc_motion_vectors_offset(const struct v4l2_ctrl_hevc_sps *sps);
>> +size_t hantro_hevc_luma_compress_offset(const struct v4l2_ctrl_hevc_sps *sps);
>> +size_t hantro_hevc_chroma_compress_offset(const struct v4l2_ctrl_hevc_sps *sps);
>>
>> static inline size_t
>> hantro_h264_mv_size(unsigned int width, unsigned int height)
> Thanks,
> Ezequiel
>
Powered by blists - more mailing lists