Hey Tomasz,
Thanks for taking the time to review this carefully!
On Thu, 2019-03-28 at 18:57 +0900, Tomasz Figa wrote:
> On Tue, Mar 5, 2019 at 4:27 AM Ezequiel Garcia <ezequiel@xxxxxxxxxxxxx> wrote:
> [snip]
> > +static const struct rockchip_vpu_fmt *
> > +rockchip_vpu_find_format(struct rockchip_vpu_ctx *ctx, u32 fourcc)
> > +{
> > + struct rockchip_vpu_dev *dev = ctx->dev;
> > + const struct rockchip_vpu_fmt *formats;
> > + unsigned int num_fmts, i;
> > +
> > + formats = dev->variant->dec_fmts;
> > + num_fmts = dev->variant->num_dec_fmts;
> > + for (i = 0; i < num_fmts; i++)
> > + if (formats[i].fourcc == fourcc)
> > + return &formats[i];
> > + return NULL;
> > +}
> > +
> > +static const struct rockchip_vpu_fmt *
> > +rockchip_vpu_get_default_fmt(struct rockchip_vpu_ctx *ctx, bool bitstream)
> > +{
> > + struct rockchip_vpu_dev *dev = ctx->dev;
> > + const struct rockchip_vpu_fmt *formats;
> > + unsigned int num_fmts, i;
> > +
> > + formats = dev->variant->dec_fmts;
> > + num_fmts = dev->variant->num_dec_fmts;
> > + for (i = 0; i < num_fmts; i++)
> > + if (bitstream == (formats[i].codec_mode != RK_VPU_MODE_NONE))
> > + return &formats[i];
> > + return NULL;
> > +}
> > +
> > +static int vidioc_querycap(struct file *file, void *priv,
> > + struct v4l2_capability *cap)
> > +{
> > + struct rockchip_vpu_dev *vpu = video_drvdata(file);
> > +
> > + strlcpy(cap->driver, vpu->dev->driver->name, sizeof(cap->driver));
> > + strlcpy(cap->card, vpu->vfd_dec->name, sizeof(cap->card));
> > + snprintf(cap->bus_info, sizeof(cap->bus_info), "platform: %s",
> > + vpu->dev->driver->name);
> > + return 0;
> > +}
> > +
> > +static int vidioc_enum_framesizes(struct file *file, void *priv,
> > + struct v4l2_frmsizeenum *fsize)
> > +{
> > + struct rockchip_vpu_ctx *ctx = fh_to_ctx(priv);
> > + const struct rockchip_vpu_fmt *fmt;
> > +
> > + if (fsize->index != 0) {
> > + vpu_debug(0, "invalid frame size index (expected 0, got %d)\n",
> > + fsize->index);
> > + return -EINVAL;
> > + }
> > +
> > + fmt = rockchip_vpu_find_format(ctx, fsize->pixel_format);
> > + if (!fmt) {
> > + vpu_debug(0, "unsupported bitstream format (%08x)\n",
> > + fsize->pixel_format);
> > + return -EINVAL;
> > + }
> > +
> > + /* This only makes sense for codec formats */
>
> "coded formats"
>
OK.
> > + if (fmt->codec_mode == RK_VPU_MODE_NONE)
> > + return -EINVAL;
> > +
> > + fsize->type = V4L2_FRMSIZE_TYPE_STEPWISE;
> > + fsize->stepwise = fmt->frmsize;
> > +
> > + return 0;
> > +}
> > +
> > +static int vidioc_enum_fmt_vid_cap_mplane(struct file *file, void *priv,
> > + struct v4l2_fmtdesc *f)
> > +{
> > + struct rockchip_vpu_dev *dev = video_drvdata(file);
> > + const struct rockchip_vpu_fmt *fmt;
> > + const struct rockchip_vpu_fmt *formats;
> > + int num_fmts, i, j = 0;
> > +
> > + formats = dev->variant->dec_fmts;
> > + num_fmts = dev->variant->num_dec_fmts;
> > + for (i = 0; i < num_fmts; i++) {
> > + /* Skip compressed formats */
> > + if (formats[i].codec_mode != RK_VPU_MODE_NONE)
> > + continue;
> > + if (j == f->index) {
> > + fmt = &formats[i];
> > + f->pixelformat = fmt->fourcc;
> > + return 0;
> > + }
> > + ++j;
> > + }
> > + return -EINVAL;
> > +}
> > +
> > +static int vidioc_enum_fmt_vid_out_mplane(struct file *file, void *priv,
> > + struct v4l2_fmtdesc *f)
> > +{
> > + struct rockchip_vpu_dev *dev = video_drvdata(file);
> > + const struct rockchip_vpu_fmt *formats;
> > + const struct rockchip_vpu_fmt *fmt;
> > + int num_fmts, i, j = 0;
> > +
> > + formats = dev->variant->dec_fmts;
> > + num_fmts = dev->variant->num_dec_fmts;
> > + for (i = 0; i < num_fmts; i++) {
> > + if (formats[i].codec_mode == RK_VPU_MODE_NONE)
> > + continue;
> > + if (j == f->index) {
> > + fmt = &formats[i];
> > + f->pixelformat = fmt->fourcc;
> > + return 0;
> > + }
> > + ++j;
> > + }
> > + return -EINVAL;
> > +}
>
> All the functions above can be trivially generalized to be used for
> both encoder and decoder. Moreover,
> vidioc_enum_fmt_vid_{cap,out}_mplane() can be trivially merged into
> one function to cover both buffer types.
>
Right.
> > +
> > +static int vidioc_g_fmt_out_mplane(struct file *file, void *priv,
> > + struct v4l2_format *f)
> > +{
> > + struct v4l2_pix_format_mplane *pix_mp = &f->fmt.pix_mp;
> > + struct rockchip_vpu_ctx *ctx = fh_to_ctx(priv);
> > +
> > + vpu_debug(4, "f->type = %d\n", f->type);
> > +
> > + *pix_mp = ctx->src_fmt;
> > +
> > + return 0;
> > +}
> > +
> > +static int vidioc_g_fmt_cap_mplane(struct file *file, void *priv,
> > + struct v4l2_format *f)
> > +{
> > + struct v4l2_pix_format_mplane *pix_mp = &f->fmt.pix_mp;
> > + struct rockchip_vpu_ctx *ctx = fh_to_ctx(priv);
> > +
> > + vpu_debug(4, "f->type = %d\n", f->type);
> > +
> > + *pix_mp = ctx->dst_fmt;
> > +
> > + return 0;
> > +}
> > +
> > +static int
> > +vidioc_try_fmt_cap_mplane(struct file *file, void *priv, struct v4l2_format *f)
> > +{
> > + struct rockchip_vpu_ctx *ctx = fh_to_ctx(priv);
> > + struct v4l2_pix_format_mplane *pix_mp = &f->fmt.pix_mp;
> > + const struct rockchip_vpu_fmt *fmt;
> > + unsigned int width, height;
> > +
> > + vpu_debug(4, "%c%c%c%c\n",
> > + (pix_mp->pixelformat & 0x7f),
> > + (pix_mp->pixelformat >> 8) & 0x7f,
> > + (pix_mp->pixelformat >> 16) & 0x7f,
> > + (pix_mp->pixelformat >> 24) & 0x7f);
> > +
> > + fmt = rockchip_vpu_find_format(ctx, pix_mp->pixelformat);
> > + if (!fmt) {
> > + fmt = rockchip_vpu_get_default_fmt(ctx, false);
> > + f->fmt.pix.pixelformat = fmt->fourcc;
> > + }
> > +
> > + pix_mp->field = V4L2_FIELD_NONE;
> > + width = clamp(pix_mp->width,
> > + ctx->vpu_src_fmt->frmsize.min_width,
> > + ctx->vpu_src_fmt->frmsize.max_width);
> > + height = clamp(pix_mp->height,
> > + ctx->vpu_src_fmt->frmsize.min_height,
> > + ctx->vpu_src_fmt->frmsize.max_height);
> > + /* Round up to macroblocks. */
> > + width = round_up(width, ctx->vpu_src_fmt->frmsize.step_width);
> > + height = round_up(height, ctx->vpu_src_fmt->frmsize.step_height);
> > +
> > + /* Fill remaining fields */
> > + v4l2_fill_pixfmt_mp(pix_mp, fmt->fourcc, width, height);
> > + return 0;
> > +}
> > +
> > +static int
> > +vidioc_try_fmt_out_mplane(struct file *file, void *priv, struct v4l2_format *f)
> > +{
> > + struct rockchip_vpu_ctx *ctx = fh_to_ctx(priv);
> > + struct v4l2_pix_format_mplane *pix_mp = &f->fmt.pix_mp;
> > + const struct rockchip_vpu_fmt *fmt;
> > +
> > + fmt = rockchip_vpu_find_format(ctx, pix_mp->pixelformat);
> > + if (!fmt) {
> > + fmt = rockchip_vpu_get_default_fmt(ctx, true);
> > + f->fmt.pix.pixelformat = fmt->fourcc;
> > + }
> > +
> > + pix_mp->num_planes = 1;
> > + pix_mp->field = V4L2_FIELD_NONE;
> > + pix_mp->width = clamp(pix_mp->width,
> > + fmt->frmsize.min_width,
> > + fmt->frmsize.max_width);
> > + pix_mp->height = clamp(pix_mp->height,
> > + fmt->frmsize.min_height,
> > + fmt->frmsize.max_height);
> > + /* Round up to macroblocks. */
> > + pix_mp->width = round_up(pix_mp->width, fmt->frmsize.step_width);
> > + pix_mp->height = round_up(pix_mp->height, fmt->frmsize.step_height);
> > +
> > + /*
> > + * For compressed formats the application can specify
> > + * sizeimage. If the application passes a zero sizeimage,
> > + * let's default to the maximum frame size.
> > + */
> > + if (!pix_mp->plane_fmt[0].sizeimage)
> > + pix_mp->plane_fmt[0].sizeimage = fmt->header_size +
> > + pix_mp->width * pix_mp->height * fmt->max_depth;
> > + return 0;
> > +}
> > +
> > +void rockchip_vpu_dec_reset_dst_fmt(struct rockchip_vpu_dev *vpu,
> > + struct rockchip_vpu_ctx *ctx)
> > +{
> > + struct v4l2_pix_format_mplane *fmt = &ctx->dst_fmt;
> > + unsigned int width, height;
> > +
> > + ctx->vpu_dst_fmt = rockchip_vpu_get_default_fmt(ctx, false);
> > +
> > + memset(fmt, 0, sizeof(*fmt));
> > +
> > + width = clamp(fmt->width, ctx->vpu_src_fmt->frmsize.min_width,
> > + ctx->vpu_src_fmt->frmsize.max_width);
> > + height = clamp(fmt->height, ctx->vpu_src_fmt->frmsize.min_height,
> > + ctx->vpu_src_fmt->frmsize.max_height);
> > + fmt->field = V4L2_FIELD_NONE;
> > + fmt->colorspace = V4L2_COLORSPACE_JPEG,
> > + fmt->ycbcr_enc = V4L2_YCBCR_ENC_DEFAULT;
> > + fmt->quantization = V4L2_QUANTIZATION_DEFAULT;
> > + fmt->xfer_func = V4L2_XFER_FUNC_DEFAULT;
> > +
> > + v4l2_fill_pixfmt_mp(fmt, ctx->vpu_dst_fmt->fourcc, width, height);
>
> Rather than duplicating most of the work already done in
> vidioc_try_fmt_*(), could we just call it here?
>
Yeah, I can try, see how it goes.
> > +}
> > +
> > +void rockchip_vpu_dec_reset_src_fmt(struct rockchip_vpu_dev *vpu,
> > + struct rockchip_vpu_ctx *ctx)
> > +{
> > + struct v4l2_pix_format_mplane *fmt = &ctx->src_fmt;
> > +
> > + ctx->vpu_src_fmt = rockchip_vpu_get_default_fmt(ctx, true);
> > +
> > + memset(fmt, 0, sizeof(*fmt));
> > +
> > + fmt->num_planes = 1;
> > + fmt->width = clamp(fmt->width, ctx->vpu_src_fmt->frmsize.min_width,
> > + ctx->vpu_src_fmt->frmsize.max_width);
> > + fmt->height = clamp(fmt->height, ctx->vpu_src_fmt->frmsize.min_height,
> > + ctx->vpu_src_fmt->frmsize.max_height);
> > + fmt->pixelformat = ctx->vpu_src_fmt->fourcc;
> > + fmt->field = V4L2_FIELD_NONE;
> > + fmt->colorspace = V4L2_COLORSPACE_JPEG,
> > + fmt->ycbcr_enc = V4L2_YCBCR_ENC_DEFAULT;
> > + fmt->quantization = V4L2_QUANTIZATION_DEFAULT;
> > + fmt->xfer_func = V4L2_XFER_FUNC_DEFAULT;
> > +
> > + fmt->plane_fmt[0].sizeimage = ctx->vpu_src_fmt->header_size +
> > + fmt->width * fmt->height * ctx->vpu_src_fmt->max_depth;
>
> Ditto.
>
> > +}
> > +
> > +static int
> > +vidioc_s_fmt_out_mplane(struct file *file, void *priv, struct v4l2_format *f)
> > +{
> > + struct v4l2_pix_format_mplane *pix_mp = &f->fmt.pix_mp;
> > + struct rockchip_vpu_ctx *ctx = fh_to_ctx(priv);
> > + struct rockchip_vpu_dev *vpu = ctx->dev;
> > + struct vb2_queue *vq, *peer_vq;
> > + int ret;
> > +
> > + /* Change not allowed if queue is streaming. */
> > + vq = v4l2_m2m_get_vq(ctx->fh.m2m_ctx, f->type);
> > + if (vb2_is_streaming(vq))
> > + return -EBUSY;
> > +
> > + /*
> > + * Since format change on the OUTPUT queue will reset
> > + * the CAPTURE queue, we can't allow doing so
> > + * when the CAPTURE queue has buffers allocated.
> > + */
> > + peer_vq = v4l2_m2m_get_vq(ctx->fh.m2m_ctx,
> > + V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE);
> > + if (vb2_is_busy(peer_vq) &&
> > + (pix_mp->pixelformat != ctx->src_fmt.pixelformat ||
> > + pix_mp->height != ctx->src_fmt.height ||
> > + pix_mp->width != ctx->src_fmt.width))
> > + return -EBUSY;
>
> I'd say that in general we don't want any S_FMT(OUT) to be allowed if
> CAP has buffers allocated, so maybe just checking vb2_is_busy(peer_vq)
> is enough?
>
Hm, this has been confusing to no end for me. This check comes from s_fmt_cap
in rockchip_vpu_enc.c, at the time we came to the conclusion that
vb2_is_busy(peer_queue) wasn't enough.
Taking a step back, the S_FMT order is first encoded format, then raw format.
So for encoders the user is expected to do first S_FMT(CAP), second S_FMT(OUT).
For decoders, first S_FMT(OUT), second S_FMT(CAP)... and that's why we are interested
on a strict vb2_is_busy check.
The question is how we came to a different conclusion on the encoder.
> > +
> > + ret = vidioc_try_fmt_out_mplane(file, priv, f);
> > + if (ret)
> > + return ret;
> > +
> > + ctx->vpu_src_fmt = rockchip_vpu_find_format(ctx, pix_mp->pixelformat);
> > + ctx->src_fmt = *pix_mp;
> > +
> > + vpu_debug(0, "OUTPUT codec mode: %d\n", ctx->vpu_src_fmt->codec_mode);
> > + vpu_debug(0, "fmt - w: %d, h: %d\n",
> > + pix_mp->width, pix_mp->height);
> > +
> > + rockchip_vpu_dec_reset_dst_fmt(vpu, ctx);
> > + return 0;
> > +}
> > +
> > +static int
> > +vidioc_s_fmt_cap_mplane(struct file *file, void *priv, struct v4l2_format *f)
> > +{
> > + struct v4l2_pix_format_mplane *pix_mp = &f->fmt.pix_mp;
> > + struct rockchip_vpu_ctx *ctx = fh_to_ctx(priv);
> > + struct vb2_queue *vq;
> > + int ret;
> > +
> > + /* Change not allowed if queue is streaming. */
> > + vq = v4l2_m2m_get_vq(ctx->fh.m2m_ctx, f->type);
> > + if (vb2_is_streaming(vq))
> > + return -EBUSY;
> > +
> > + ret = vidioc_try_fmt_cap_mplane(file, priv, f);
> > + if (ret)
> > + return ret;
> > +
> > + ctx->vpu_dst_fmt = rockchip_vpu_find_format(ctx, pix_mp->pixelformat);
> > + ctx->dst_fmt = *pix_mp;
> > +
> > + vpu_debug(0, "CAPTURE codec mode: %d\n", ctx->vpu_dst_fmt->codec_mode);
> > + vpu_debug(0, "fmt - w: %d, h: %d\n",
> > + pix_mp->width, pix_mp->height);
> > + return 0;
> > +}
> > +
> > +const struct v4l2_ioctl_ops rockchip_vpu_dec_ioctl_ops = {
> > + .vidioc_querycap = vidioc_querycap,
> > + .vidioc_enum_framesizes = vidioc_enum_framesizes,
> > +
> > + .vidioc_try_fmt_vid_cap_mplane = vidioc_try_fmt_cap_mplane,
> > + .vidioc_try_fmt_vid_out_mplane = vidioc_try_fmt_out_mplane,
> > + .vidioc_s_fmt_vid_out_mplane = vidioc_s_fmt_out_mplane,
> > + .vidioc_s_fmt_vid_cap_mplane = vidioc_s_fmt_cap_mplane,
> > + .vidioc_g_fmt_vid_out_mplane = vidioc_g_fmt_out_mplane,
> > + .vidioc_g_fmt_vid_cap_mplane = vidioc_g_fmt_cap_mplane,
> > + .vidioc_enum_fmt_vid_out_mplane = vidioc_enum_fmt_vid_out_mplane,
> > + .vidioc_enum_fmt_vid_cap_mplane = vidioc_enum_fmt_vid_cap_mplane,
> > +
> > + .vidioc_reqbufs = v4l2_m2m_ioctl_reqbufs,
> > + .vidioc_querybuf = v4l2_m2m_ioctl_querybuf,
> > + .vidioc_qbuf = v4l2_m2m_ioctl_qbuf,
> > + .vidioc_dqbuf = v4l2_m2m_ioctl_dqbuf,
> > + .vidioc_prepare_buf = v4l2_m2m_ioctl_prepare_buf,
> > + .vidioc_create_bufs = v4l2_m2m_ioctl_create_bufs,
> > + .vidioc_expbuf = v4l2_m2m_ioctl_expbuf,
> > +
> > + .vidioc_subscribe_event = v4l2_ctrl_subscribe_event,
> > + .vidioc_unsubscribe_event = v4l2_event_unsubscribe,
> > +
> > + .vidioc_streamon = v4l2_m2m_ioctl_streamon,
> > + .vidioc_streamoff = v4l2_m2m_ioctl_streamoff,
> > +};
> > +
> > +static int rockchip_vpu_queue_setup(struct vb2_queue *vq,
> > + unsigned int *num_buffers,
> > + unsigned int *num_planes,
> > + unsigned int sizes[],
> > + struct device *alloc_devs[])
> > +{
> > + struct rockchip_vpu_ctx *ctx = vb2_get_drv_priv(vq);
> > + struct v4l2_pix_format_mplane *pixfmt;
> > + int i;
> > +
> > + switch (vq->type) {
> > + case V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE:
> > + pixfmt = &ctx->dst_fmt;
> > + break;
> > + case V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE:
> > + pixfmt = &ctx->src_fmt;
> > + break;
> > + default:
> > + vpu_err("invalid queue type: %d\n", vq->type);
> > + return -EINVAL;
> > + }
>
> Hmm, would it make sense to have separate vb2_queue_ops for each queue
> and create helper functions, which take the necessary queue-specific
> data as arguments, if appropriate?
>
Yup, sounds good.
> > +
> > + if (*num_planes) {
> > + if (*num_planes != pixfmt->num_planes)
> > + return -EINVAL;
> > + for (i = 0; i < pixfmt->num_planes; ++i)
> > + if (sizes[i] < pixfmt->plane_fmt[i].sizeimage)
> > + return -EINVAL;
> > + return 0;
> > + }
> > +
> > + *num_planes = pixfmt->num_planes;
> > + for (i = 0; i < pixfmt->num_planes; ++i)
> > + sizes[i] = pixfmt->plane_fmt[i].sizeimage;
> > + return 0;
> > +}
> > +
> > +static int rockchip_vpu_buf_prepare(struct vb2_buffer *vb)
> > +{
> > + struct vb2_v4l2_buffer *vbuf = to_vb2_v4l2_buffer(vb);
> > + struct vb2_queue *vq = vb->vb2_queue;
> > + struct rockchip_vpu_ctx *ctx = vb2_get_drv_priv(vq);
> > + const struct rockchip_vpu_fmt *vpu_fmt;
> > + struct v4l2_pix_format_mplane *pixfmt;
> > + unsigned int sz;
> > + int ret = 0;
> > + int i;
> > +
> > + switch (vq->type) {
> > + case V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE:
> > + vpu_fmt = ctx->vpu_dst_fmt;
> > + pixfmt = &ctx->dst_fmt;
> > + break;
> > + case V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE:
> > + vpu_fmt = ctx->vpu_src_fmt;
> > + pixfmt = &ctx->src_fmt;
> > +
> > + if (vbuf->field == V4L2_FIELD_ANY)
> > + vbuf->field = V4L2_FIELD_NONE;
> > + if (vbuf->field != V4L2_FIELD_NONE) {
> > + vpu_debug(4, "field %d not supported\n",
> > + vbuf->field);
> > + return -EINVAL;
> > + }
> > + break;
> > + default:
> > + vpu_err("invalid queue type: %d\n", vq->type);
> > + return -EINVAL;
> > + }
> > +
> > + for (i = 0; i < pixfmt->num_planes; ++i) {
> > + sz = pixfmt->plane_fmt[i].sizeimage;
> > + vpu_debug(4, "plane %d size: %ld, sizeimage: %u\n",
> > + i, vb2_plane_size(vb, i), sz);
> > + if (vb2_plane_size(vb, i) < sz) {
> > + vpu_err("plane %d is too small for output\n", i);
> > + ret = -EINVAL;
> > + break;
> > + }
> > + }
> > +
> > + return ret;
> > +}
> > +
> > +static void rockchip_vpu_buf_queue(struct vb2_buffer *vb)
> > +{
> > + struct rockchip_vpu_ctx *ctx = vb2_get_drv_priv(vb->vb2_queue);
> > + struct vb2_v4l2_buffer *vbuf = to_vb2_v4l2_buffer(vb);
> > +
> > + v4l2_m2m_buf_queue(ctx->fh.m2m_ctx, vbuf);
> > +}
> > +
> > +static int rockchip_vpu_start_streaming(struct vb2_queue *q, unsigned int count)
> > +{
> > + struct rockchip_vpu_ctx *ctx = vb2_get_drv_priv(q);
> > + enum rockchip_vpu_codec_mode codec_mode;
> > + int ret = 0;
> > +
> > + if (V4L2_TYPE_IS_OUTPUT(q->type))
> > + ctx->sequence_out = 0;
> > + else
> > + ctx->sequence_cap = 0;
> > +
> > + codec_mode = ctx->vpu_src_fmt->codec_mode;
> > +
> > + vpu_debug(4, "Codec mode = %d\n", codec_mode);
> > + ctx->codec_ops = &ctx->dev->variant->codec_ops[codec_mode];
>
> Codec mode should be a function of the OUT queue, so I think this
> should be also under if (V4L2_TYPE_IS_OUTPUT()). This would make most
> of the code in this function specific to the OUT queue and possibly a
> good argument for the comment about having separate vb2 queue ops I
> posted above.
>
Yup.
> > +
> > + if (V4L2_TYPE_IS_OUTPUT(q->type))
> > + if (ctx->codec_ops && ctx->codec_ops->start)
> > + ret = ctx->codec_ops->start(ctx);
> > + return ret;
> > +}
> > +
> > +static void rockchip_vpu_stop_streaming(struct vb2_queue *q)
> > +{
> > + struct rockchip_vpu_ctx *ctx = vb2_get_drv_priv(q);
> > +
> > + if (V4L2_TYPE_IS_OUTPUT(q->type))
> > + if (ctx->codec_ops && ctx->codec_ops->stop)
> > + ctx->codec_ops->stop(ctx);
> > +
> > + /* The mem2mem framework calls v4l2_m2m_cancel_job before
>
> CodingStyle: Multiline comments should start with one empty line.
>
Oops. Seems I have a tendency for this!
> > + * .stop_streaming, so there isn't any job running and
> > + * it is safe to return all the buffers.
> > + */
> > + for (;;) {
> > + struct vb2_v4l2_buffer *vbuf;
> > +
> > + if (V4L2_TYPE_IS_OUTPUT(q->type))
> > + vbuf = v4l2_m2m_src_buf_remove(ctx->fh.m2m_ctx);
> > + else
> > + vbuf = v4l2_m2m_dst_buf_remove(ctx->fh.m2m_ctx);
> > + if (!vbuf)
> > + break;
> > + v4l2_ctrl_request_complete(vbuf->vb2_buf.req_obj.req, &ctx->ctrl_handler);
> > + v4l2_m2m_buf_done(vbuf, VB2_BUF_STATE_ERROR);
> > + }
> > +}
> > +
> > +static void rockchip_vpu_buf_request_complete(struct vb2_buffer *vb)
> > +{
> > + struct rockchip_vpu_ctx *ctx = vb2_get_drv_priv(vb->vb2_queue);
> > +
> > + v4l2_ctrl_request_complete(vb->req_obj.req, &ctx->ctrl_handler);
> > +}
> > +
> > +static int rockchip_vpu_buf_out_validate(struct vb2_buffer *vb)
> > +{
> > + struct vb2_v4l2_buffer *vbuf = to_vb2_v4l2_buffer(vb);
> > +
> > + vbuf->field = V4L2_FIELD_NONE;
>
> Similar comment as in earlier patch. Is validate expected to correct
> the values or just check?
>
See Hans' reply.
> > + return 0;
> > +}
> > +
> > +const struct vb2_ops rockchip_vpu_dec_queue_ops = {
> > + .queue_setup = rockchip_vpu_queue_setup,
> > + .buf_prepare = rockchip_vpu_buf_prepare,
> > + .buf_queue = rockchip_vpu_buf_queue,
> > + .buf_out_validate = rockchip_vpu_buf_out_validate,
> > + .buf_request_complete = rockchip_vpu_buf_request_complete,
> > + .start_streaming = rockchip_vpu_start_streaming,
> > + .stop_streaming = rockchip_vpu_stop_streaming,
> > + .wait_prepare = vb2_ops_wait_prepare,
> > + .wait_finish = vb2_ops_wait_finish,
> > +};
> > diff --git a/drivers/staging/media/rockchip/vpu/rockchip_vpu_drv.c b/drivers/staging/media/rockchip/vpu/rockchip_vpu_drv.c
> > index 937e9cfb4568..27a9da86f1d0 100644
> > --- a/drivers/staging/media/rockchip/vpu/rockchip_vpu_drv.c
> > +++ b/drivers/staging/media/rockchip/vpu/rockchip_vpu_drv.c
> > @@ -36,6 +36,11 @@ module_param_named(debug, rockchip_vpu_debug, int, 0644);
> > MODULE_PARM_DESC(debug,
> > "Debug level - higher value produces more verbose messages");
> >
> > +enum rockchip_vpu_type {
> > + RK_VPU_ENCODER,
> > + RK_VPU_DECODER,
> > +};
>
> Given that we already have ctx->is_enc, any reason not to have this
> represented in the same way everywhere?
>
The reason is that I'd rather have:
rockchip_vpu_video_device_register(vpu, RK_VPU_ENCODER);
Than:
rockchip_vpu_video_device_register(vpu, true);
We could have a vpu_type field instead of that is_enc boolean.
I think the code would be even cleaner.
> > +
> > void *rockchip_vpu_get_ctrl(struct rockchip_vpu_ctx *ctx, u32 id)
> > {
> > struct v4l2_ctrl *ctrl;
> > @@ -81,7 +86,11 @@ static void rockchip_vpu_job_finish(struct rockchip_vpu_dev *vpu,
> >
> > avail_size = vb2_plane_size(&dst->vb2_buf, 0) -
> > ctx->vpu_dst_fmt->header_size;
> > - if (bytesused <= avail_size) {
> > + /* For decoders set bytesused as per the output picture. */
> > + if (!ctx->is_enc) {
> > + dst->vb2_buf.planes[0].bytesused =
> > + ctx->dst_fmt.plane_fmt[0].sizeimage;
> > + } else if (bytesused <= avail_size) {
> > /*
> > * This works while JPEG is the only encoder this driver
> > * supports. We will have to abstract this step, or get
>
> avail_size is specific for the encoder. How about clearly separating
> decoder and encoder code with a separate if? It would introduce one
> more level of nesting, but IMHO it would be more readable, due to
> better separation of code paths. (Or going one step further, encoder
> and decoder-specific code could be moved to separate functions.)
>
Sounds good!
> if (ctx->is_enc) {
> size_t avail_size = ...;
> // encoder code
> if (bytesused <= avail_size) {
> //
> } else {
> //
> }
> } else {
> // decoder code
> }
>
> > @@ -159,6 +168,44 @@ static struct v4l2_m2m_ops vpu_m2m_ops = {
> > .device_run = device_run,
> > };
> >
> > +static int
> > +dec_queue_init(void *priv, struct vb2_queue *src_vq, struct vb2_queue *dst_vq)
> > +{
> > + struct rockchip_vpu_ctx *ctx = priv;
> > + int ret;
> > +
> > + src_vq->type = V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE;
> > + src_vq->io_modes = VB2_MMAP | VB2_DMABUF;
> > + src_vq->drv_priv = ctx;
> > + src_vq->ops = &rockchip_vpu_dec_queue_ops;
> > + src_vq->mem_ops = &vb2_dma_contig_memops;
> > + src_vq->dma_attrs = DMA_ATTR_ALLOC_SINGLE_PAGES |
> > + DMA_ATTR_NO_KERNEL_MAPPING;
> > + src_vq->buf_struct_size = sizeof(struct v4l2_m2m_buffer);
> > + src_vq->timestamp_flags = V4L2_BUF_FLAG_TIMESTAMP_COPY;
> > + src_vq->lock = &ctx->dev->vpu_mutex;
> > + src_vq->dev = ctx->dev->v4l2_dev.dev;
> > + src_vq->supports_requests = true;
> > +
> > + ret = vb2_queue_init(src_vq);
> > + if (ret)
> > + return ret;
> > +
> > + dst_vq->type = V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE;
> > + dst_vq->io_modes = VB2_MMAP | VB2_DMABUF;
> > + dst_vq->drv_priv = ctx;
> > + dst_vq->bidirectional = true;
> > + dst_vq->ops = &rockchip_vpu_dec_queue_ops;
> > + dst_vq->mem_ops = &vb2_dma_contig_memops;
> > + dst_vq->dma_attrs = DMA_ATTR_ALLOC_SINGLE_PAGES;
>
> Any need for kernel mapping of decoder output?
>
Nope, I just missed this.
Thanks a lot for the detailed review!
Eze
