All of lore.kernel.org
 help / color / mirror / Atom feed
From: Dafna Hirschfeld <dafna.hirschfeld@collabora.com>
To: Alexandre Courbot <acourbot@chromium.org>
Cc: Tiffany Lin <tiffany.lin@mediatek.com>,
	Andrew-CT Chen <andrew-ct.chen@mediatek.com>,
	Hans Verkuil <hverkuil-cisco@xs4all.nl>,
	Yunfei Dong <yunfei.dong@mediatek.com>,
	Mauro Carvalho Chehab <mchehab@kernel.org>,
	linux-media@vger.kernel.org, linux-kernel@vger.kernel.org,
	linux-mediatek@lists.infradead.org,
	Hsin-Yi Wang <hsinyi@chromium.org>
Subject: Re: [PATCH v5 05/14] media: mtk-vcodec: venc: support START and STOP commands
Date: Fri, 28 May 2021 10:03:27 +0300	[thread overview]
Message-ID: <9b37044d-f909-9169-3d22-fa6c5f788822@collabora.com> (raw)
In-Reply-To: <20210519143011.1175546-6-acourbot@chromium.org>

Hi,

I applied this patchset and tested the stateful encoder on debian with the command:

[gst-master] root@debian:~/gst-build# gst-launch-1.0 filesrc location=images/jelly-800-640.YU12 ! rawvideoparse width=800 height=640 format=i420 ! videoconvert ! v4l2h264enc ! h264parse ! mp4mux ! filesink location=jelly-800-640.mp4

I get:

Setting pipeline[   79.703879] [MTK_V4L2] level=0 fops_vcodec_open(),190: encoder capability 10000000
  to PAUSED ...
Pipeline is PREROLLING ...
Redistribute latency...
[   80.621076] mtk-iommu 10205000.iommu: Partial TLB flush timed out, falling back to full flush
[   80.631232] mtk-iommu 10205000.iommu: Partial TLB flush timed out, falling back to full flush
[   80.640878] mtk-iommu 10205000.iommu: Partial TLB flush timed out, falling back to full flush
[   80.650766] mtk-iommu 10205000.iommu: Partial TLB flush timed out, falling back to full flush
[   80.660430] mtk-iommu 10205000.iommu: Partial TLB flush timed out, falling back to full flush
[   80.670194] mtk-iommu 10205000.iommu: Partial TLB flush timed out, falling back to full flush
[   80.680967] mtk-iommu 10205000.iommu: Partial TLB flush timed out, falling back to full flush
[   80.691376] mtk-iommu 10205000.iommu: Partial TLB flush timed out, falling back to full flush
[   80.701718] mtk-iommu 10205000.iommu: Partial TLB flush timed out, falling back to full flush
[   80.712106] mtk-iommu 10205000.iommu: Partial TLB flush timed out, falling back to full flush
[   80.722272] [MTK_V4L2] level=0 mtk_venc_set_param(),371: fmt 0x3, P/L 0/0, w/h 800/640, buf 800/640, fps/bps 25/4000000, gop 0, i_period 0
Pipeline is PREROLLED ...
Setting pipeline to PLAYING ...
New clock: GstSystemClock
[   81.918747] [MTK_V4L2][ERROR] mtk_vcodec_wait_for_done_ctx:32: [3] ctx->type=1, cmd=1, wait_event_interruptible_timeout time=1000ms out 0 0!
[   81.931392] [MTK_VCODEC][ERROR][3]: h264_encode_frame() irq_status=0 failed
[   81.938470] [MTK_V4L2][ERROR] mtk_venc_worker:1219: venc_if_encode failed=-5
[   82.974746] [MTK_V4L2][ERROR] mtk_vcodec_wait_for_done_ctx:32: [3] ctx->type=1, cmd=1, wait_event_interruptible_timeout time=1000ms out 0 0!
[   82.987392] [MTK_VCODEC][ERROR][3]: h264_encode_frame() irq_status=0 failed
[   82.994471] [MTK_V4L2][ERROR] mtk_venc_worker:1219: venc_if_encode failed=-5
[  104.163977] cros-ec-dev cros-ec-dev.2.auto: Some logs may have been dropped...
0:00:00.4 / 99:99:99.
0:00:00.4 / 99:99:99.
0:00:00.4 / 99:99:99.
0:00:00.4 / 99:99:99.
0:00:00.4 / 99:99:99.
0:00:00.4 / 99:99:99.
0:00:00.4 / 99:99:99.
0:00:00.4 / 99:99:99.
0:00:00.4 / 99:99:99.
0:00:00.4 / 99:99:99.
^Chandling interrupt.

And then the streaming hangs. The same error happens without this patchset, but without
this patchset the statful encoder does not support V4L2_ENC_CMD_STOP/START needed by the spec.
I am not sure what cause the error and wether those mtk-iommu erros has to do with that. The issue
could also come from the mtk-vpu used by the encoder.
Do you have any idea where this can come from?

Thanks,
Dafna



On 19.05.21 17:30, Alexandre Courbot wrote:
> The V4L2 encoder specification requires encoders to support the
> V4L2_ENC_CMD_START and V4L2_ENC_CMD_STOP commands. Add support for these
> to the mtk-vcodec encoder by reusing the same flush buffer as used by
> the decoder driver.
> 
> Signed-off-by: Alexandre Courbot <acourbot@chromium.org>
> [hsinyi: fix double-free issue if flush buffer was not dequeued by the
> time streamoff is called]
> Signed-off-by: Hsin-Yi Wang <hsinyi@chromium.org>
> ---
>   .../platform/mtk-vcodec/mtk_vcodec_drv.h      |   2 +
>   .../platform/mtk-vcodec/mtk_vcodec_enc.c      | 135 +++++++++++++++++-
>   .../platform/mtk-vcodec/mtk_vcodec_enc_drv.c  |   4 +
>   3 files changed, 134 insertions(+), 7 deletions(-)
> 
> diff --git a/drivers/media/platform/mtk-vcodec/mtk_vcodec_drv.h b/drivers/media/platform/mtk-vcodec/mtk_vcodec_drv.h
> index c6fe61253f43..37a62c0f406f 100644
> --- a/drivers/media/platform/mtk-vcodec/mtk_vcodec_drv.h
> +++ b/drivers/media/platform/mtk-vcodec/mtk_vcodec_drv.h
> @@ -252,6 +252,7 @@ struct vdec_pic_info {
>    * @last_decoded_picinfo: pic information get from latest decode
>    * @empty_flush_buf: a fake size-0 capture buffer that indicates flush. Only
>    *		     to be used with encoder and stateful decoder.
> + * @is_flushing: set to true if flushing is in progress.
>    *
>    * @colorspace: enum v4l2_colorspace; supplemental to pixelformat
>    * @ycbcr_enc: enum v4l2_ycbcr_encoding, Y'CbCr encoding
> @@ -290,6 +291,7 @@ struct mtk_vcodec_ctx {
>   	struct work_struct encode_work;
>   	struct vdec_pic_info last_decoded_picinfo;
>   	struct v4l2_m2m_buffer empty_flush_buf;
> +	bool is_flushing;
>   
>   	enum v4l2_colorspace colorspace;
>   	enum v4l2_ycbcr_encoding ycbcr_enc;
> diff --git a/drivers/media/platform/mtk-vcodec/mtk_vcodec_enc.c b/drivers/media/platform/mtk-vcodec/mtk_vcodec_enc.c
> index 4831052f475d..4701dea251ca 100644
> --- a/drivers/media/platform/mtk-vcodec/mtk_vcodec_enc.c
> +++ b/drivers/media/platform/mtk-vcodec/mtk_vcodec_enc.c
> @@ -659,6 +659,7 @@ static int vidioc_venc_dqbuf(struct file *file, void *priv,
>   			     struct v4l2_buffer *buf)
>   {
>   	struct mtk_vcodec_ctx *ctx = fh_to_ctx(priv);
> +	int ret;
>   
>   	if (ctx->state == MTK_STATE_ABORT) {
>   		mtk_v4l2_err("[%d] Call on QBUF after unrecoverable error",
> @@ -666,7 +667,77 @@ static int vidioc_venc_dqbuf(struct file *file, void *priv,
>   		return -EIO;
>   	}
>   
> -	return v4l2_m2m_dqbuf(file, ctx->m2m_ctx, buf);
> +	ret = v4l2_m2m_dqbuf(file, ctx->m2m_ctx, buf);
> +	if (ret)
> +		return ret;
> +
> +	/*
> +	 * Complete flush if the user dequeued the 0-payload LAST buffer.
> +	 * We check the payload because a buffer with the LAST flag can also
> +	 * be seen during resolution changes. If we happen to be flushing at
> +	 * that time, the last buffer before the resolution changes could be
> +	 * misinterpreted for the buffer generated by the flush and terminate
> +	 * it earlier than we want.
> +	 */
> +	if (!V4L2_TYPE_IS_OUTPUT(buf->type) &&
> +	    buf->flags & V4L2_BUF_FLAG_LAST &&
> +	    buf->m.planes[0].bytesused == 0 &&
> +	    ctx->is_flushing) {
> +		/*
> +		 * Last CAPTURE buffer is dequeued, we can allow another flush
> +		 * to take place.
> +		 */
> +		ctx->is_flushing = false;
> +	}
> +
> +	return 0;
> +}
> +
> +static int vidioc_encoder_cmd(struct file *file, void *priv,
> +			      struct v4l2_encoder_cmd *cmd)
> +{
> +	struct mtk_vcodec_ctx *ctx = fh_to_ctx(priv);
> +	struct vb2_queue *src_vq, *dst_vq;
> +	int ret;
> +
> +	ret = v4l2_m2m_ioctl_try_encoder_cmd(file, priv, cmd);
> +	if (ret)
> +		return ret;
> +
> +	/* Calling START or STOP is invalid if a flush is in progress */
> +	if (ctx->is_flushing)
> +		return -EBUSY;
> +
> +	mtk_v4l2_debug(1, "encoder cmd=%u", cmd->cmd);
> +
> +	dst_vq = v4l2_m2m_get_vq(ctx->m2m_ctx,
> +				V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE);
> +	switch (cmd->cmd) {
> +	case V4L2_ENC_CMD_STOP:
> +		src_vq = v4l2_m2m_get_vq(ctx->m2m_ctx,
> +				V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE);
> +		if (!vb2_is_streaming(src_vq)) {
> +			mtk_v4l2_debug(1, "Output stream is off. No need to flush.");
> +			return 0;
> +		}
> +		if (!vb2_is_streaming(dst_vq)) {
> +			mtk_v4l2_debug(1, "Capture stream is off. No need to flush.");
> +			return 0;
> +		}
> +		ctx->is_flushing = true;
> +		v4l2_m2m_buf_queue(ctx->m2m_ctx, &ctx->empty_flush_buf.vb);
> +		v4l2_m2m_try_schedule(ctx->m2m_ctx);
> +		break;
> +
> +	case V4L2_ENC_CMD_START:
> +		vb2_clear_last_buffer_dequeued(dst_vq);
> +		break;
> +
> +	default:
> +		return -EINVAL;
> +	}
> +
> +	return 0;
>   }
>   
>   const struct v4l2_ioctl_ops mtk_venc_ioctl_ops = {
> @@ -702,6 +773,9 @@ const struct v4l2_ioctl_ops mtk_venc_ioctl_ops = {
>   
>   	.vidioc_g_selection		= vidioc_venc_g_selection,
>   	.vidioc_s_selection		= vidioc_venc_s_selection,
> +
> +	.vidioc_encoder_cmd		= vidioc_encoder_cmd,
> +	.vidioc_try_encoder_cmd		= v4l2_m2m_ioctl_try_encoder_cmd,
>   };
>   
>   static int vb2ops_venc_queue_setup(struct vb2_queue *vq,
> @@ -869,9 +943,39 @@ static void vb2ops_venc_stop_streaming(struct vb2_queue *q)
>   			dst_buf->vb2_buf.planes[0].bytesused = 0;
>   			v4l2_m2m_buf_done(dst_buf, VB2_BUF_STATE_ERROR);
>   		}
> +		/* STREAMOFF on the CAPTURE queue completes any ongoing flush */
> +		if (ctx->is_flushing) {
> +			struct v4l2_m2m_buffer *b, *n;
> +
> +			mtk_v4l2_debug(1, "STREAMOFF called while flushing");
> +			/*
> +			 * STREAMOFF could be called before the flush buffer is
> +			 * dequeued. Check whether empty flush buf is still in
> +			 * queue before removing it.
> +			 */
> +			v4l2_m2m_for_each_src_buf_safe(ctx->m2m_ctx, b, n) {
> +				if (b == &ctx->empty_flush_buf) {
> +					v4l2_m2m_src_buf_remove_by_buf(
> +							ctx->m2m_ctx, &b->vb);
> +					break;
> +				}
> +			}
> +			ctx->is_flushing = false;
> +		}
>   	} else {
> -		while ((src_buf = v4l2_m2m_src_buf_remove(ctx->m2m_ctx)))
> -			v4l2_m2m_buf_done(src_buf, VB2_BUF_STATE_ERROR);
> +		while ((src_buf = v4l2_m2m_src_buf_remove(ctx->m2m_ctx))) {
> +			if (src_buf != &ctx->empty_flush_buf.vb)
> +				v4l2_m2m_buf_done(src_buf, VB2_BUF_STATE_ERROR);
> +		}
> +		if (ctx->is_flushing) {
> +			/*
> +			 * If we are in the middle of a flush, put the flush
> +			 * buffer back into the queue so the next CAPTURE
> +			 * buffer gets returned with the LAST flag set.
> +			 */
> +			v4l2_m2m_buf_queue(ctx->m2m_ctx,
> +					   &ctx->empty_flush_buf.vb);
> +		}
>   	}
>   
>   	if ((q->type == V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE &&
> @@ -971,12 +1075,15 @@ static int mtk_venc_param_change(struct mtk_vcodec_ctx *ctx)
>   {
>   	struct venc_enc_param enc_prm;
>   	struct vb2_v4l2_buffer *vb2_v4l2 = v4l2_m2m_next_src_buf(ctx->m2m_ctx);
> -	struct mtk_video_enc_buf *mtk_buf =
> -			container_of(vb2_v4l2, struct mtk_video_enc_buf,
> -				     m2m_buf.vb);
> -
> +	struct mtk_video_enc_buf *mtk_buf;
>   	int ret = 0;
>   
> +	/* Don't upcast the empty flush buffer */
> +	if (vb2_v4l2 == &ctx->empty_flush_buf.vb)
> +		return 0;
> +
> +	mtk_buf = container_of(vb2_v4l2, struct mtk_video_enc_buf, m2m_buf.vb);
> +
>   	memset(&enc_prm, 0, sizeof(enc_prm));
>   	if (mtk_buf->param_change == MTK_ENCODE_PARAM_NONE)
>   		return 0;
> @@ -1062,6 +1169,20 @@ static void mtk_venc_worker(struct work_struct *work)
>   	}
>   
>   	src_buf = v4l2_m2m_src_buf_remove(ctx->m2m_ctx);
> +
> +	/*
> +	 * If we see the flush buffer, send an empty buffer with the LAST flag
> +	 * to the client. is_flushing will be reset at the time the buffer
> +	 * is dequeued.
> +	 */
> +	if (src_buf == &ctx->empty_flush_buf.vb) {
> +		vb2_set_plane_payload(&dst_buf->vb2_buf, 0, 0);
> +		dst_buf->flags |= V4L2_BUF_FLAG_LAST;
> +		v4l2_m2m_buf_done(dst_buf, VB2_BUF_STATE_DONE);
> +		v4l2_m2m_job_finish(ctx->dev->m2m_dev_enc, ctx->m2m_ctx);
> +		return;
> +	}
> +
>   	memset(&frm_buf, 0, sizeof(frm_buf));
>   	for (i = 0; i < src_buf->vb2_buf.num_planes ; i++) {
>   		frm_buf.fb_addr[i].dma_addr =
> diff --git a/drivers/media/platform/mtk-vcodec/mtk_vcodec_enc_drv.c b/drivers/media/platform/mtk-vcodec/mtk_vcodec_enc_drv.c
> index 7d7b8cfc2cc5..2dd6fef896df 100644
> --- a/drivers/media/platform/mtk-vcodec/mtk_vcodec_enc_drv.c
> +++ b/drivers/media/platform/mtk-vcodec/mtk_vcodec_enc_drv.c
> @@ -131,6 +131,7 @@ static int fops_vcodec_open(struct file *file)
>   	struct mtk_vcodec_dev *dev = video_drvdata(file);
>   	struct mtk_vcodec_ctx *ctx = NULL;
>   	int ret = 0;
> +	struct vb2_queue *src_vq;
>   
>   	ctx = kzalloc(sizeof(*ctx), GFP_KERNEL);
>   	if (!ctx)
> @@ -164,6 +165,9 @@ static int fops_vcodec_open(struct file *file)
>   				ret);
>   		goto err_m2m_ctx_init;
>   	}
> +	src_vq = v4l2_m2m_get_vq(ctx->m2m_ctx,
> +				V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE);
> +	ctx->empty_flush_buf.vb.vb2_buf.vb2_queue = src_vq;
>   	mtk_vcodec_enc_set_default_params(ctx);
>   
>   	if (v4l2_fh_is_singular(&ctx->fh)) {
> 

WARNING: multiple messages have this Message-ID
From: Dafna Hirschfeld <dafna.hirschfeld@collabora.com>
To: Alexandre Courbot <acourbot@chromium.org>
Cc: Tiffany Lin <tiffany.lin@mediatek.com>,
	Andrew-CT Chen <andrew-ct.chen@mediatek.com>,
	Hans Verkuil <hverkuil-cisco@xs4all.nl>,
	Yunfei Dong <yunfei.dong@mediatek.com>,
	Mauro Carvalho Chehab <mchehab@kernel.org>,
	linux-media@vger.kernel.org, linux-kernel@vger.kernel.org,
	linux-mediatek@lists.infradead.org,
	Hsin-Yi Wang <hsinyi@chromium.org>
Subject: Re: [PATCH v5 05/14] media: mtk-vcodec: venc: support START and STOP commands
Date: Fri, 28 May 2021 10:03:27 +0300	[thread overview]
Message-ID: <9b37044d-f909-9169-3d22-fa6c5f788822@collabora.com> (raw)
In-Reply-To: <20210519143011.1175546-6-acourbot@chromium.org>

Hi,

I applied this patchset and tested the stateful encoder on debian with the command:

[gst-master] root@debian:~/gst-build# gst-launch-1.0 filesrc location=images/jelly-800-640.YU12 ! rawvideoparse width=800 height=640 format=i420 ! videoconvert ! v4l2h264enc ! h264parse ! mp4mux ! filesink location=jelly-800-640.mp4

I get:

Setting pipeline[   79.703879] [MTK_V4L2] level=0 fops_vcodec_open(),190: encoder capability 10000000
  to PAUSED ...
Pipeline is PREROLLING ...
Redistribute latency...
[   80.621076] mtk-iommu 10205000.iommu: Partial TLB flush timed out, falling back to full flush
[   80.631232] mtk-iommu 10205000.iommu: Partial TLB flush timed out, falling back to full flush
[   80.640878] mtk-iommu 10205000.iommu: Partial TLB flush timed out, falling back to full flush
[   80.650766] mtk-iommu 10205000.iommu: Partial TLB flush timed out, falling back to full flush
[   80.660430] mtk-iommu 10205000.iommu: Partial TLB flush timed out, falling back to full flush
[   80.670194] mtk-iommu 10205000.iommu: Partial TLB flush timed out, falling back to full flush
[   80.680967] mtk-iommu 10205000.iommu: Partial TLB flush timed out, falling back to full flush
[   80.691376] mtk-iommu 10205000.iommu: Partial TLB flush timed out, falling back to full flush
[   80.701718] mtk-iommu 10205000.iommu: Partial TLB flush timed out, falling back to full flush
[   80.712106] mtk-iommu 10205000.iommu: Partial TLB flush timed out, falling back to full flush
[   80.722272] [MTK_V4L2] level=0 mtk_venc_set_param(),371: fmt 0x3, P/L 0/0, w/h 800/640, buf 800/640, fps/bps 25/4000000, gop 0, i_period 0
Pipeline is PREROLLED ...
Setting pipeline to PLAYING ...
New clock: GstSystemClock
[   81.918747] [MTK_V4L2][ERROR] mtk_vcodec_wait_for_done_ctx:32: [3] ctx->type=1, cmd=1, wait_event_interruptible_timeout time=1000ms out 0 0!
[   81.931392] [MTK_VCODEC][ERROR][3]: h264_encode_frame() irq_status=0 failed
[   81.938470] [MTK_V4L2][ERROR] mtk_venc_worker:1219: venc_if_encode failed=-5
[   82.974746] [MTK_V4L2][ERROR] mtk_vcodec_wait_for_done_ctx:32: [3] ctx->type=1, cmd=1, wait_event_interruptible_timeout time=1000ms out 0 0!
[   82.987392] [MTK_VCODEC][ERROR][3]: h264_encode_frame() irq_status=0 failed
[   82.994471] [MTK_V4L2][ERROR] mtk_venc_worker:1219: venc_if_encode failed=-5
[  104.163977] cros-ec-dev cros-ec-dev.2.auto: Some logs may have been dropped...
0:00:00.4 / 99:99:99.
0:00:00.4 / 99:99:99.
0:00:00.4 / 99:99:99.
0:00:00.4 / 99:99:99.
0:00:00.4 / 99:99:99.
0:00:00.4 / 99:99:99.
0:00:00.4 / 99:99:99.
0:00:00.4 / 99:99:99.
0:00:00.4 / 99:99:99.
0:00:00.4 / 99:99:99.
^Chandling interrupt.

And then the streaming hangs. The same error happens without this patchset, but without
this patchset the statful encoder does not support V4L2_ENC_CMD_STOP/START needed by the spec.
I am not sure what cause the error and wether those mtk-iommu erros has to do with that. The issue
could also come from the mtk-vpu used by the encoder.
Do you have any idea where this can come from?

Thanks,
Dafna



On 19.05.21 17:30, Alexandre Courbot wrote:
> The V4L2 encoder specification requires encoders to support the
> V4L2_ENC_CMD_START and V4L2_ENC_CMD_STOP commands. Add support for these
> to the mtk-vcodec encoder by reusing the same flush buffer as used by
> the decoder driver.
> 
> Signed-off-by: Alexandre Courbot <acourbot@chromium.org>
> [hsinyi: fix double-free issue if flush buffer was not dequeued by the
> time streamoff is called]
> Signed-off-by: Hsin-Yi Wang <hsinyi@chromium.org>
> ---
>   .../platform/mtk-vcodec/mtk_vcodec_drv.h      |   2 +
>   .../platform/mtk-vcodec/mtk_vcodec_enc.c      | 135 +++++++++++++++++-
>   .../platform/mtk-vcodec/mtk_vcodec_enc_drv.c  |   4 +
>   3 files changed, 134 insertions(+), 7 deletions(-)
> 
> diff --git a/drivers/media/platform/mtk-vcodec/mtk_vcodec_drv.h b/drivers/media/platform/mtk-vcodec/mtk_vcodec_drv.h
> index c6fe61253f43..37a62c0f406f 100644
> --- a/drivers/media/platform/mtk-vcodec/mtk_vcodec_drv.h
> +++ b/drivers/media/platform/mtk-vcodec/mtk_vcodec_drv.h
> @@ -252,6 +252,7 @@ struct vdec_pic_info {
>    * @last_decoded_picinfo: pic information get from latest decode
>    * @empty_flush_buf: a fake size-0 capture buffer that indicates flush. Only
>    *		     to be used with encoder and stateful decoder.
> + * @is_flushing: set to true if flushing is in progress.
>    *
>    * @colorspace: enum v4l2_colorspace; supplemental to pixelformat
>    * @ycbcr_enc: enum v4l2_ycbcr_encoding, Y'CbCr encoding
> @@ -290,6 +291,7 @@ struct mtk_vcodec_ctx {
>   	struct work_struct encode_work;
>   	struct vdec_pic_info last_decoded_picinfo;
>   	struct v4l2_m2m_buffer empty_flush_buf;
> +	bool is_flushing;
>   
>   	enum v4l2_colorspace colorspace;
>   	enum v4l2_ycbcr_encoding ycbcr_enc;
> diff --git a/drivers/media/platform/mtk-vcodec/mtk_vcodec_enc.c b/drivers/media/platform/mtk-vcodec/mtk_vcodec_enc.c
> index 4831052f475d..4701dea251ca 100644
> --- a/drivers/media/platform/mtk-vcodec/mtk_vcodec_enc.c
> +++ b/drivers/media/platform/mtk-vcodec/mtk_vcodec_enc.c
> @@ -659,6 +659,7 @@ static int vidioc_venc_dqbuf(struct file *file, void *priv,
>   			     struct v4l2_buffer *buf)
>   {
>   	struct mtk_vcodec_ctx *ctx = fh_to_ctx(priv);
> +	int ret;
>   
>   	if (ctx->state == MTK_STATE_ABORT) {
>   		mtk_v4l2_err("[%d] Call on QBUF after unrecoverable error",
> @@ -666,7 +667,77 @@ static int vidioc_venc_dqbuf(struct file *file, void *priv,
>   		return -EIO;
>   	}
>   
> -	return v4l2_m2m_dqbuf(file, ctx->m2m_ctx, buf);
> +	ret = v4l2_m2m_dqbuf(file, ctx->m2m_ctx, buf);
> +	if (ret)
> +		return ret;
> +
> +	/*
> +	 * Complete flush if the user dequeued the 0-payload LAST buffer.
> +	 * We check the payload because a buffer with the LAST flag can also
> +	 * be seen during resolution changes. If we happen to be flushing at
> +	 * that time, the last buffer before the resolution changes could be
> +	 * misinterpreted for the buffer generated by the flush and terminate
> +	 * it earlier than we want.
> +	 */
> +	if (!V4L2_TYPE_IS_OUTPUT(buf->type) &&
> +	    buf->flags & V4L2_BUF_FLAG_LAST &&
> +	    buf->m.planes[0].bytesused == 0 &&
> +	    ctx->is_flushing) {
> +		/*
> +		 * Last CAPTURE buffer is dequeued, we can allow another flush
> +		 * to take place.
> +		 */
> +		ctx->is_flushing = false;
> +	}
> +
> +	return 0;
> +}
> +
> +static int vidioc_encoder_cmd(struct file *file, void *priv,
> +			      struct v4l2_encoder_cmd *cmd)
> +{
> +	struct mtk_vcodec_ctx *ctx = fh_to_ctx(priv);
> +	struct vb2_queue *src_vq, *dst_vq;
> +	int ret;
> +
> +	ret = v4l2_m2m_ioctl_try_encoder_cmd(file, priv, cmd);
> +	if (ret)
> +		return ret;
> +
> +	/* Calling START or STOP is invalid if a flush is in progress */
> +	if (ctx->is_flushing)
> +		return -EBUSY;
> +
> +	mtk_v4l2_debug(1, "encoder cmd=%u", cmd->cmd);
> +
> +	dst_vq = v4l2_m2m_get_vq(ctx->m2m_ctx,
> +				V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE);
> +	switch (cmd->cmd) {
> +	case V4L2_ENC_CMD_STOP:
> +		src_vq = v4l2_m2m_get_vq(ctx->m2m_ctx,
> +				V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE);
> +		if (!vb2_is_streaming(src_vq)) {
> +			mtk_v4l2_debug(1, "Output stream is off. No need to flush.");
> +			return 0;
> +		}
> +		if (!vb2_is_streaming(dst_vq)) {
> +			mtk_v4l2_debug(1, "Capture stream is off. No need to flush.");
> +			return 0;
> +		}
> +		ctx->is_flushing = true;
> +		v4l2_m2m_buf_queue(ctx->m2m_ctx, &ctx->empty_flush_buf.vb);
> +		v4l2_m2m_try_schedule(ctx->m2m_ctx);
> +		break;
> +
> +	case V4L2_ENC_CMD_START:
> +		vb2_clear_last_buffer_dequeued(dst_vq);
> +		break;
> +
> +	default:
> +		return -EINVAL;
> +	}
> +
> +	return 0;
>   }
>   
>   const struct v4l2_ioctl_ops mtk_venc_ioctl_ops = {
> @@ -702,6 +773,9 @@ const struct v4l2_ioctl_ops mtk_venc_ioctl_ops = {
>   
>   	.vidioc_g_selection		= vidioc_venc_g_selection,
>   	.vidioc_s_selection		= vidioc_venc_s_selection,
> +
> +	.vidioc_encoder_cmd		= vidioc_encoder_cmd,
> +	.vidioc_try_encoder_cmd		= v4l2_m2m_ioctl_try_encoder_cmd,
>   };
>   
>   static int vb2ops_venc_queue_setup(struct vb2_queue *vq,
> @@ -869,9 +943,39 @@ static void vb2ops_venc_stop_streaming(struct vb2_queue *q)
>   			dst_buf->vb2_buf.planes[0].bytesused = 0;
>   			v4l2_m2m_buf_done(dst_buf, VB2_BUF_STATE_ERROR);
>   		}
> +		/* STREAMOFF on the CAPTURE queue completes any ongoing flush */
> +		if (ctx->is_flushing) {
> +			struct v4l2_m2m_buffer *b, *n;
> +
> +			mtk_v4l2_debug(1, "STREAMOFF called while flushing");
> +			/*
> +			 * STREAMOFF could be called before the flush buffer is
> +			 * dequeued. Check whether empty flush buf is still in
> +			 * queue before removing it.
> +			 */
> +			v4l2_m2m_for_each_src_buf_safe(ctx->m2m_ctx, b, n) {
> +				if (b == &ctx->empty_flush_buf) {
> +					v4l2_m2m_src_buf_remove_by_buf(
> +							ctx->m2m_ctx, &b->vb);
> +					break;
> +				}
> +			}
> +			ctx->is_flushing = false;
> +		}
>   	} else {
> -		while ((src_buf = v4l2_m2m_src_buf_remove(ctx->m2m_ctx)))
> -			v4l2_m2m_buf_done(src_buf, VB2_BUF_STATE_ERROR);
> +		while ((src_buf = v4l2_m2m_src_buf_remove(ctx->m2m_ctx))) {
> +			if (src_buf != &ctx->empty_flush_buf.vb)
> +				v4l2_m2m_buf_done(src_buf, VB2_BUF_STATE_ERROR);
> +		}
> +		if (ctx->is_flushing) {
> +			/*
> +			 * If we are in the middle of a flush, put the flush
> +			 * buffer back into the queue so the next CAPTURE
> +			 * buffer gets returned with the LAST flag set.
> +			 */
> +			v4l2_m2m_buf_queue(ctx->m2m_ctx,
> +					   &ctx->empty_flush_buf.vb);
> +		}
>   	}
>   
>   	if ((q->type == V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE &&
> @@ -971,12 +1075,15 @@ static int mtk_venc_param_change(struct mtk_vcodec_ctx *ctx)
>   {
>   	struct venc_enc_param enc_prm;
>   	struct vb2_v4l2_buffer *vb2_v4l2 = v4l2_m2m_next_src_buf(ctx->m2m_ctx);
> -	struct mtk_video_enc_buf *mtk_buf =
> -			container_of(vb2_v4l2, struct mtk_video_enc_buf,
> -				     m2m_buf.vb);
> -
> +	struct mtk_video_enc_buf *mtk_buf;
>   	int ret = 0;
>   
> +	/* Don't upcast the empty flush buffer */
> +	if (vb2_v4l2 == &ctx->empty_flush_buf.vb)
> +		return 0;
> +
> +	mtk_buf = container_of(vb2_v4l2, struct mtk_video_enc_buf, m2m_buf.vb);
> +
>   	memset(&enc_prm, 0, sizeof(enc_prm));
>   	if (mtk_buf->param_change == MTK_ENCODE_PARAM_NONE)
>   		return 0;
> @@ -1062,6 +1169,20 @@ static void mtk_venc_worker(struct work_struct *work)
>   	}
>   
>   	src_buf = v4l2_m2m_src_buf_remove(ctx->m2m_ctx);
> +
> +	/*
> +	 * If we see the flush buffer, send an empty buffer with the LAST flag
> +	 * to the client. is_flushing will be reset at the time the buffer
> +	 * is dequeued.
> +	 */
> +	if (src_buf == &ctx->empty_flush_buf.vb) {
> +		vb2_set_plane_payload(&dst_buf->vb2_buf, 0, 0);
> +		dst_buf->flags |= V4L2_BUF_FLAG_LAST;
> +		v4l2_m2m_buf_done(dst_buf, VB2_BUF_STATE_DONE);
> +		v4l2_m2m_job_finish(ctx->dev->m2m_dev_enc, ctx->m2m_ctx);
> +		return;
> +	}
> +
>   	memset(&frm_buf, 0, sizeof(frm_buf));
>   	for (i = 0; i < src_buf->vb2_buf.num_planes ; i++) {
>   		frm_buf.fb_addr[i].dma_addr =
> diff --git a/drivers/media/platform/mtk-vcodec/mtk_vcodec_enc_drv.c b/drivers/media/platform/mtk-vcodec/mtk_vcodec_enc_drv.c
> index 7d7b8cfc2cc5..2dd6fef896df 100644
> --- a/drivers/media/platform/mtk-vcodec/mtk_vcodec_enc_drv.c
> +++ b/drivers/media/platform/mtk-vcodec/mtk_vcodec_enc_drv.c
> @@ -131,6 +131,7 @@ static int fops_vcodec_open(struct file *file)
>   	struct mtk_vcodec_dev *dev = video_drvdata(file);
>   	struct mtk_vcodec_ctx *ctx = NULL;
>   	int ret = 0;
> +	struct vb2_queue *src_vq;
>   
>   	ctx = kzalloc(sizeof(*ctx), GFP_KERNEL);
>   	if (!ctx)
> @@ -164,6 +165,9 @@ static int fops_vcodec_open(struct file *file)
>   				ret);
>   		goto err_m2m_ctx_init;
>   	}
> +	src_vq = v4l2_m2m_get_vq(ctx->m2m_ctx,
> +				V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE);
> +	ctx->empty_flush_buf.vb.vb2_buf.vb2_queue = src_vq;
>   	mtk_vcodec_enc_set_default_params(ctx);
>   
>   	if (v4l2_fh_is_singular(&ctx->fh)) {
> 

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

  parent reply	other threads:[~2021-05-28  7:03 UTC|newest]

Thread overview: 78+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-05-19 14:29 [PATCH v5 00/14] media: mtk-vcodec: support for MT8183 decoder Alexandre Courbot
2021-05-19 14:29 ` Alexandre Courbot
2021-05-19 14:29 ` [PATCH v5 01/14] media: mtk-vcodec: vdec: Support H264 profile control Alexandre Courbot
2021-05-19 14:29   ` Alexandre Courbot
2021-05-21 13:36   ` Tzung-Bi Shih
2021-05-21 13:36     ` Tzung-Bi Shih
2021-05-19 14:29 ` [PATCH v5 02/14] media: mtk-vcodec: vdec: use helpers in VIDIOC_(TRY_)DECODER_CMD Alexandre Courbot
2021-05-19 14:29   ` Alexandre Courbot
2021-05-21 13:37   ` Tzung-Bi Shih
2021-05-21 13:37     ` Tzung-Bi Shih
2021-05-19 14:30 ` [PATCH v5 03/14] media: mtk-vcodec: vdec: clamp OUTPUT resolution to hardware limits Alexandre Courbot
2021-05-19 14:30   ` Alexandre Courbot
2021-05-21 13:37   ` Tzung-Bi Shih
2021-05-21 13:37     ` Tzung-Bi Shih
2021-05-19 14:30 ` [PATCH v5 04/14] media: mtk-vcodec: make flush buffer reusable by encoder Alexandre Courbot
2021-05-19 14:30   ` Alexandre Courbot
2021-05-21 13:37   ` Tzung-Bi Shih
2021-05-21 13:37     ` Tzung-Bi Shih
2021-05-19 14:30 ` [PATCH v5 05/14] media: mtk-vcodec: venc: support START and STOP commands Alexandre Courbot
2021-05-19 14:30   ` Alexandre Courbot
2021-05-21 13:37   ` Tzung-Bi Shih
2021-05-21 13:37     ` Tzung-Bi Shih
2021-05-27 10:10     ` Alexandre Courbot
2021-05-27 10:10       ` Alexandre Courbot
2021-05-28  7:03   ` Dafna Hirschfeld [this message]
2021-05-28  7:03     ` Dafna Hirschfeld
2021-05-28  7:43     ` Dafna Hirschfeld
2021-05-28  7:43       ` Dafna Hirschfeld
2021-07-05  5:04       ` Alexandre Courbot
2021-07-05  5:04         ` Alexandre Courbot
2021-07-05  5:04     ` Alexandre Courbot
2021-07-05  5:04       ` Alexandre Courbot
2021-07-06 15:17       ` Enric Balletbo Serra
2021-07-06 15:17         ` Enric Balletbo Serra
2021-05-19 14:30 ` [PATCH v5 06/14] media: mtk-vcodec: vdec: move stateful ops into their own file Alexandre Courbot
2021-05-19 14:30   ` Alexandre Courbot
2021-05-21 13:37   ` Tzung-Bi Shih
2021-05-21 13:37     ` Tzung-Bi Shih
2021-05-27 10:10     ` Alexandre Courbot
2021-05-27 10:10       ` Alexandre Courbot
2021-05-19 14:30 ` [PATCH v5 07/14] media: mtk-vcodec: vdec: handle firmware version field Alexandre Courbot
2021-05-19 14:30   ` Alexandre Courbot
2021-05-21 13:37   ` Tzung-Bi Shih
2021-05-21 13:37     ` Tzung-Bi Shih
2021-05-19 14:30 ` [PATCH v5 08/14] media: mtk-vcodec: support version 2 of decoder firmware ABI Alexandre Courbot
2021-05-19 14:30   ` Alexandre Courbot
2021-05-21 13:37   ` Tzung-Bi Shih
2021-05-21 13:37     ` Tzung-Bi Shih
2021-05-19 14:30 ` [PATCH v5 09/14] media: add Mediatek's MM21 format Alexandre Courbot
2021-05-19 14:30   ` Alexandre Courbot
2021-05-21 13:37   ` Tzung-Bi Shih
2021-05-21 13:37     ` Tzung-Bi Shih
2021-05-19 14:30 ` [PATCH v5 10/14] media: mtk-vcodec: vdec: support stateless API Alexandre Courbot
2021-05-19 14:30   ` Alexandre Courbot
2021-05-21 13:38   ` Tzung-Bi Shih
2021-05-21 13:38     ` Tzung-Bi Shih
2021-05-19 14:30 ` [PATCH v5 11/14] media: mtk-vcodec: vdec: support stateless H.264 decoding Alexandre Courbot
2021-05-19 14:30   ` Alexandre Courbot
2021-05-21 13:38   ` Tzung-Bi Shih
2021-05-21 13:38     ` Tzung-Bi Shih
2021-05-27 10:10     ` Alexandre Courbot
2021-05-27 10:10       ` Alexandre Courbot
2021-05-19 14:30 ` [PATCH v5 12/14] media: mtk-vcodec: vdec: add media device if using stateless api Alexandre Courbot
2021-05-19 14:30   ` Alexandre Courbot
2021-05-21 13:38   ` Tzung-Bi Shih
2021-05-21 13:38     ` Tzung-Bi Shih
2021-05-19 14:30 ` [PATCH v5 13/14] dt-bindings: media: document mediatek,mt8183-vcodec-dec Alexandre Courbot
2021-05-19 14:30   ` [PATCH v5 13/14] dt-bindings: media: document mediatek, mt8183-vcodec-dec Alexandre Courbot
2021-05-19 14:30 ` [PATCH v5 14/14] media: mtk-vcodec: enable MT8183 decoder Alexandre Courbot
2021-05-19 14:30   ` Alexandre Courbot
2021-05-21 13:38   ` Tzung-Bi Shih
2021-05-21 13:38     ` Tzung-Bi Shih
2021-05-27  8:08 ` [PATCH v5 00/14] media: mtk-vcodec: support for " Hans Verkuil
2021-05-27  8:08   ` Hans Verkuil
2021-05-27 10:10   ` Alexandre Courbot
2021-05-27 10:10     ` Alexandre Courbot
2021-05-27 10:18     ` Hans Verkuil
2021-05-27 10:18       ` Hans Verkuil

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=9b37044d-f909-9169-3d22-fa6c5f788822@collabora.com \
    --to=dafna.hirschfeld@collabora.com \
    --cc=acourbot@chromium.org \
    --cc=andrew-ct.chen@mediatek.com \
    --cc=hsinyi@chromium.org \
    --cc=hverkuil-cisco@xs4all.nl \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-media@vger.kernel.org \
    --cc=linux-mediatek@lists.infradead.org \
    --cc=mchehab@kernel.org \
    --cc=tiffany.lin@mediatek.com \
    --cc=yunfei.dong@mediatek.com \
    --subject='Re: [PATCH v5 05/14] media: mtk-vcodec: venc: support START and STOP commands' \
    /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

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.