All of lore.kernel.org
 help / color / mirror / Atom feed
From: javier Martin <javier.martin@vista-silicon.com>
To: Philipp Zabel <p.zabel@pengutronix.de>
Cc: linux-media@vger.kernel.org,
	sakari.ailus@maxwell.research.nokia.com,
	kyungmin.park@samsung.com, s.nawrocki@samsung.com,
	laurent.pinchart@ideasonboard.com, mchehab@infradead.org,
	s.hauer@pengutronix.de
Subject: Re: [PATCH v3] media: coda: Add driver for Coda video codec.
Date: Wed, 18 Jul 2012 09:12:21 +0200	[thread overview]
Message-ID: <CACKLOr3rOPgwMCRdj3ARR+0655Qp=BfEXq0TsB7TU-hO4NSsqg@mail.gmail.com> (raw)
In-Reply-To: <1342459273.2535.665.camel@pizza.hi.pengutronix.de>

Hi Pilipp,

On 16 July 2012 19:21, Philipp Zabel <p.zabel@pengutronix.de> wrote:
> Hi Javier,
>
> I'm trying to use this driver (on i.MX53) with GStreamer, with two
> buffers on the encoded bitstream side. I'm getting negative bytesused
> values calculated in coda_isr().
>
> The CODA_REG_BIT_WR_PTR address, which is initialized in
> coda_start_streaming() only moves around in the first buffer (so far
> attempts to set it from device_run() have failed). This causes negative
> bytesused values to be calculated in coda_isr() if the second buffer
> (which has a higher BB_START address than WR_PTR) is active.
> I'll repeat my observations inline below.
>
> Are you using just one buffer for encoded frames? Does your firmware
> behave differently?

Apparently, it does; see an example where I am using two frames:

get_frame: (0) bytesused = 3579
get_frame: (1) bytesused = 3834
get_frame: (0) bytesused = 14925
get_frame: (1) bytesused = 4493
get_frame: (0) bytesused = 3977
get_frame: (1) bytesused = 3855
get_frame: (0) bytesused = 3853
get_frame: (1) bytesused = 3646
get_frame: (0) bytesused = 3797
get_frame: (1) bytesused = 3695
get_frame: (0) bytesused = 3653
get_frame: (1) bytesused = 3770
get_frame: (0) bytesused = 3418
get_frame: (1) bytesused = 3749
get_frame: (0) bytesused = 3621
get_frame: (1) bytesused = 3743
get_frame: (0) bytesused = 3506
get_frame: (1) bytesused = 3523
get_frame: (0) bytesused = 14844

As you can see I don't get any negative values. Furthermore, the
values seem rather correct since as gop_size is 16. There are 15
smaller P frames between two larger I frames.

> I see there is a comment about the expected register setting not working
> for CODA_REG_BIT_STREAM_CTRL in start_streaming(). Could this be
> related?

I don't think so. This means that the following line:

coda_write(dev, (3 << 3), CODA_REG_BIT_STREAM_CTRL);

should be:

coda_write(dev, (CODADX6_STREAM_BUF_PIC_RESET |
CODADX6_STREAM_BUF_PIC_FLUSH), CODA_REG_BIT_STREAM_CTRL);

But the latter does not work.

> Also, I've missed two problems with platform device removal and module
> autoloading before, see below.

Fine.

> Am Donnerstag, den 12.07.2012, 09:11 +0200 schrieb Javier Martin:
>> Coda is a range of video codecs from Chips&Media that
>> support H.264, H.263, MPEG4 and other video standards.
>>
>> Currently only support for the codadx6 included in the
>> i.MX27 SoC is added. H.264 and MPEG4 video encoding
>> are the only supported capabilities by now.
>>
>> Signed-off-by: Javier Martin <javier.martin@vista-silicon.com>
>> ---
>> Changes since v2:
>>  - Make hex constants lower case.
>>  - Fix bytesperline.
>>  - Make some functions and variables static.
>>  - Fix logical &&
>>  - Remove get_qops()
>> ---
> [...]
>> +static int vidioc_g_fmt(struct coda_ctx *ctx, struct v4l2_format *f)
>> +{
>> +     struct vb2_queue *vq;
>> +     struct coda_q_data *q_data;
>> +
>> +     vq = v4l2_m2m_get_vq(ctx->m2m_ctx, f->type);
>> +     if (!vq)
>> +             return -EINVAL;
>> +
>> +     q_data = get_q_data(ctx, f->type);
>> +
>> +     f->fmt.pix.field        = V4L2_FIELD_NONE;
>> +     f->fmt.pix.pixelformat  = q_data->fmt->fourcc;
>> +     if (f->fmt.pix.pixelformat == V4L2_PIX_FMT_YUV420) {
>> +             f->fmt.pix.width        = q_data->width;
>> +             f->fmt.pix.height       = q_data->height;
>> +             f->fmt.pix.bytesperline = round_up(f->fmt.pix.width, 2);
>> +     } else { /* encoded formats h.264/mpeg4 */
>> +             f->fmt.pix.width        = 0;
>> +             f->fmt.pix.height       = 0;
>> +             f->fmt.pix.bytesperline = 0;
>> +     }
>> +     f->fmt.pix.sizeimage    = q_data->sizeimage;
>> +
>> +     return 0;
>> +}
>> +
>> +static int vidioc_g_fmt_vid_out(struct file *file, void *priv,
>> +                             struct v4l2_format *f)
>> +{
>> +     return vidioc_g_fmt(fh_to_ctx(priv), f);
>> +}
>> +
>> +static int vidioc_g_fmt_vid_cap(struct file *file, void *priv,
>> +                             struct v4l2_format *f)
>> +{
>> +     return vidioc_g_fmt(fh_to_ctx(priv), f);
>> +}
>> +
>> +static int vidioc_try_fmt(struct coda_dev *dev, struct v4l2_format *f)
>> +{
>> +     enum v4l2_field field;
>> +
>> +     if (!find_format(dev, f))
>> +             return -EINVAL;
>> +
>> +     field = f->fmt.pix.field;
>> +     if (field == V4L2_FIELD_ANY)
>> +             field = V4L2_FIELD_NONE;
>> +     else if (V4L2_FIELD_NONE != field)
>> +             return -EINVAL;
>> +
>> +     /* V4L2 specification suggests the driver corrects the format struct
>> +      * if any of the dimensions is unsupported */
>> +     f->fmt.pix.field = field;
>> +
>> +     if (f->fmt.pix.pixelformat == V4L2_PIX_FMT_YUV420) {
>> +             v4l_bound_align_image(&f->fmt.pix.width, MIN_W, MAX_W,
>> +                                   W_ALIGN, &f->fmt.pix.height,
>> +                                   MIN_H, MAX_H, H_ALIGN, S_ALIGN);
>> +             f->fmt.pix.bytesperline = round_up(f->fmt.pix.width, 2);
>> +             f->fmt.pix.sizeimage = f->fmt.pix.height *
>> +                                     f->fmt.pix.bytesperline;
>> +     } else { /*encoded formats h.264/mpeg4 */
>> +             f->fmt.pix.bytesperline = CODA_MAX_FRAME_SIZE;
>
> I failed to point this out specifically, but I think if G_FMT returns
> zero in bytesperline, so should TRY_FMT.

Of course, this is my fault too. I'll fix it in a new patch.

>> +             f->fmt.pix.sizeimage = CODA_MAX_FRAME_SIZE;
>> +     }
>> +
>> +     return 0;
>> +}
>> +
>> +static int vidioc_try_fmt_vid_cap(struct file *file, void *priv,
>> +                               struct v4l2_format *f)
>> +{
>> +     struct coda_fmt *fmt;
>> +     struct coda_ctx *ctx = fh_to_ctx(priv);
>> +
>> +     fmt = find_format(ctx->dev, f);
>> +     /*
>> +      * Since decoding support is not implemented yet do not allow
>> +      * CODA_FMT_RAW formats in the capture interface.
>> +      */
>> +     if (!fmt || !(fmt->type == CODA_FMT_ENC)) {
>> +             v4l2_err(&ctx->dev->v4l2_dev,
>> +                      "Fourcc format (0x%08x) invalid.\n",
>> +                      f->fmt.pix.pixelformat);
>> +             return -EINVAL;
>> +     }
>> +
>> +     return vidioc_try_fmt(ctx->dev, f);
>> +}
> [...]
>> +/*
>> + * Mem-to-mem operations.
>> + */
>> +
>> +static int coda_isr(struct coda_dev *dev)
>> +{
>> +     struct coda_ctx *ctx;
>> +     struct vb2_buffer *src_buf, *dst_buf, *tmp_buf;
>> +     u32 wr_ptr, start_ptr;
>> +
>> +     ctx = v4l2_m2m_get_curr_priv(dev->m2m_dev);
>> +     if (ctx == NULL) {
>> +             v4l2_err(&dev->v4l2_dev, "Instance released before the end of transaction\n");
>> +             return IRQ_HANDLED;
>> +     }
>> +
>> +     if (ctx->aborting) {
>> +             v4l2_dbg(1, coda_debug, &ctx->dev->v4l2_dev,
>> +                      "task has been aborted\n");
>> +             return IRQ_HANDLED;
>> +     }
>> +
>> +     if (coda_isbusy(ctx->dev)) {
>> +             v4l2_dbg(1, coda_debug, &ctx->dev->v4l2_dev,
>> +                      "coda is still busy!!!!\n");
>> +             return IRQ_NONE;
>> +     }
>> +
>> +     src_buf = v4l2_m2m_next_src_buf(ctx->m2m_ctx);
>> +     dst_buf = v4l2_m2m_next_dst_buf(ctx->m2m_ctx);
>> +
>> +     /* Get results from the coda */
>> +     coda_read(dev, CODA_RET_ENC_PIC_TYPE);
>> +     start_ptr = coda_read(dev, CODA_CMD_ENC_PIC_BB_START);
>> +     wr_ptr = coda_read(dev, CODA_REG_BIT_WR_PTR(ctx->idx));
>
> Here CODA_REG_BIT_WR_PTR is read, and it always contains a pointer into
> the first destination buffer that was set during start_streaming()
> below, even if I try to set it before starting the PIC_RUN command in
> device_run().
>
> start_ptr contains the correct buffer address as set in device_run().
>

I don't get the same behaviour in i.MX27 with firmware 2.2.5. See this log:

coda coda-imx27.0: (seq=0, idx=0) wr_ptr=0xa2603bf9,
start_ptr=0xa2600015 bytesused=15353
coda coda-imx27.0: (seq=1, idx=1) wr_ptr=0xa2700fd9,
start_ptr=0xa2700000 bytesused=4057
coda coda-imx27.0: (seq=2, idx=0) wr_ptr=0xa2600e62,
start_ptr=0xa2600000 bytesused=3682
coda coda-imx27.0: (seq=3, idx=1) wr_ptr=0xa2700e32,
start_ptr=0xa2700000 bytesused=3634
coda coda-imx27.0: (seq=4, idx=0) wr_ptr=0xa2600c8e,
start_ptr=0xa2600000 bytesused=3214
...

>> +     /* Calculate bytesused field */
>> +     if (dst_buf->v4l2_buf.sequence == 0) {
>> +             dst_buf->v4l2_planes[0].bytesused = (wr_ptr - start_ptr) +
>> +                                             ctx->vpu_header_size[0] +
>> +                                             ctx->vpu_header_size[1] +
>> +                                             ctx->vpu_header_size[2];
>> +     } else {
>> +             dst_buf->v4l2_planes[0].bytesused = (wr_ptr - start_ptr);
>
> So it is possible to have wr_ptr < start_ptr here.
>

Not in i.MX27.

>> +     }
>> +
>> +     v4l2_dbg(1, coda_debug, &ctx->dev->v4l2_dev, "frame size = %u\n",
>> +              wr_ptr - start_ptr);
>> +
>> +     coda_read(dev, CODA_RET_ENC_PIC_SLICE_NUM);
>> +     coda_read(dev, CODA_RET_ENC_PIC_FLAG);
>> +
>> +     if (src_buf->v4l2_buf.flags & V4L2_BUF_FLAG_KEYFRAME) {
>> +             dst_buf->v4l2_buf.flags |= V4L2_BUF_FLAG_KEYFRAME;
>> +             dst_buf->v4l2_buf.flags &= ~V4L2_BUF_FLAG_PFRAME;
>> +     } else {
>> +             dst_buf->v4l2_buf.flags |= V4L2_BUF_FLAG_PFRAME;
>> +             dst_buf->v4l2_buf.flags &= ~V4L2_BUF_FLAG_KEYFRAME;
>> +     }
>> +
>> +     /* Free previous reference picture if available */
>> +     if (ctx->reference) {
>> +             v4l2_m2m_buf_done(ctx->reference, VB2_BUF_STATE_DONE);
>> +             ctx->reference = NULL;
>> +     }
>> +
>> +     /*
>> +      * For the last frame of the gop we don't need to save
>> +      * a reference picture.
>> +      */
>> +     v4l2_m2m_dst_buf_remove(ctx->m2m_ctx);
>> +     tmp_buf = v4l2_m2m_src_buf_remove(ctx->m2m_ctx);
>> +     if (ctx->gopcounter == 0)
>> +             v4l2_m2m_buf_done(src_buf, VB2_BUF_STATE_DONE);
>> +     else
>> +             ctx->reference = tmp_buf;
>> +
>> +     v4l2_m2m_buf_done(dst_buf, VB2_BUF_STATE_DONE);
>> +
>> +     ctx->gopcounter--;
>> +     if (ctx->gopcounter < 0)
>> +             ctx->gopcounter = ctx->params.gop_size - 1;
>> +
>> +     v4l2_dbg(1, coda_debug, &dev->v4l2_dev,
>> +             "job finished: encoding frame (%d) (%s)\n",
>> +             dst_buf->v4l2_buf.sequence,
>> +             (dst_buf->v4l2_buf.flags & V4L2_BUF_FLAG_KEYFRAME) ?
>> +             "KEYFRAME" : "PFRAME");
>> +
>> +     v4l2_m2m_job_finish(ctx->dev->m2m_dev, ctx->m2m_ctx);
>> +
>> +     return IRQ_HANDLED;
>> +}
>> +
>> +static void coda_device_run(void *m2m_priv)
>> +{
>> +     struct coda_ctx *ctx = m2m_priv;
>> +     struct coda_q_data *q_data_src, *q_data_dst;
>> +     struct vb2_buffer *src_buf, *dst_buf;
>> +     struct coda_dev *dev = ctx->dev;
>> +     int force_ipicture;
>> +     int quant_param = 0;
>> +     u32 picture_y, picture_cb, picture_cr;
>> +     u32 pic_stream_buffer_addr, pic_stream_buffer_size;
>> +     u32 dst_fourcc;
>> +
>> +     src_buf = v4l2_m2m_next_src_buf(ctx->m2m_ctx);
>> +     dst_buf = v4l2_m2m_next_dst_buf(ctx->m2m_ctx);
>> +     q_data_src = get_q_data(ctx, V4L2_BUF_TYPE_VIDEO_OUTPUT);
>> +     q_data_dst = get_q_data(ctx, V4L2_BUF_TYPE_VIDEO_CAPTURE);
>> +     dst_fourcc = q_data_dst->fmt->fourcc;
>> +
>> +     src_buf->v4l2_buf.sequence = ctx->isequence;
>> +     dst_buf->v4l2_buf.sequence = ctx->isequence;
>> +     ctx->isequence++;
>> +
>> +     /*
>> +      * Workaround coda firmware BUG that only marks the first
>> +      * frame as IDR. This is a problem for some decoders that can't
>> +      * recover when a frame is lost.
>> +      */
>> +     if (src_buf->v4l2_buf.sequence % ctx->params.gop_size) {
>> +             src_buf->v4l2_buf.flags |= V4L2_BUF_FLAG_PFRAME;
>> +             src_buf->v4l2_buf.flags &= ~V4L2_BUF_FLAG_KEYFRAME;
>> +     } else {
>> +             src_buf->v4l2_buf.flags |= V4L2_BUF_FLAG_KEYFRAME;
>> +             src_buf->v4l2_buf.flags &= ~V4L2_BUF_FLAG_PFRAME;
>> +     }
>> +
>> +     /*
>> +      * Copy headers at the beginning of the first frame for H.264 only.
>> +      * In MPEG4 they are already copied by the coda.
>> +      */
>> +     if (src_buf->v4l2_buf.sequence == 0) {
>> +             pic_stream_buffer_addr =
>> +                     vb2_dma_contig_plane_dma_addr(dst_buf, 0) +
>> +                     ctx->vpu_header_size[0] +
>> +                     ctx->vpu_header_size[1] +
>> +                     ctx->vpu_header_size[2];
>> +             pic_stream_buffer_size = CODA_MAX_FRAME_SIZE -
>> +                     ctx->vpu_header_size[0] -
>> +                     ctx->vpu_header_size[1] -
>> +                     ctx->vpu_header_size[2];
>> +             memcpy(vb2_plane_vaddr(dst_buf, 0),
>> +                    &ctx->vpu_header[0][0], ctx->vpu_header_size[0]);
>> +             memcpy(vb2_plane_vaddr(dst_buf, 0) + ctx->vpu_header_size[0],
>> +                    &ctx->vpu_header[1][0], ctx->vpu_header_size[1]);
>> +             memcpy(vb2_plane_vaddr(dst_buf, 0) + ctx->vpu_header_size[0] +
>> +                     ctx->vpu_header_size[1], &ctx->vpu_header[2][0],
>> +                     ctx->vpu_header_size[2]);
>> +     } else {
>> +             pic_stream_buffer_addr =
>> +                     vb2_dma_contig_plane_dma_addr(dst_buf, 0);
>> +             pic_stream_buffer_size = CODA_MAX_FRAME_SIZE;
>> +     }
>> +
>> +     if (src_buf->v4l2_buf.flags & V4L2_BUF_FLAG_KEYFRAME) {
>> +             force_ipicture = 1;
>> +             switch (dst_fourcc) {
>> +             case V4L2_PIX_FMT_H264:
>> +                     quant_param = ctx->params.h264_intra_qp;
>> +                     break;
>> +             case V4L2_PIX_FMT_MPEG4:
>> +                     quant_param = ctx->params.mpeg4_intra_qp;
>> +                     break;
>> +             default:
>> +                     v4l2_warn(&ctx->dev->v4l2_dev,
>> +                             "cannot set intra qp, fmt not supported\n");
>> +                     break;
>> +             }
>> +     } else {
>> +             force_ipicture = 0;
>> +             switch (dst_fourcc) {
>> +             case V4L2_PIX_FMT_H264:
>> +                     quant_param = ctx->params.h264_inter_qp;
>> +                     break;
>> +             case V4L2_PIX_FMT_MPEG4:
>> +                     quant_param = ctx->params.mpeg4_inter_qp;
>> +                     break;
>> +             default:
>> +                     v4l2_warn(&ctx->dev->v4l2_dev,
>> +                             "cannot set inter qp, fmt not supported\n");
>> +                     break;
>> +             }
>> +     }
>> +
>> +     /* submit */
>> +     coda_write(dev, 0, CODA_CMD_ENC_PIC_ROT_MODE);
>> +     coda_write(dev, quant_param, CODA_CMD_ENC_PIC_QS);
>> +
>> +
>> +     picture_y = vb2_dma_contig_plane_dma_addr(src_buf, 0);
>> +     picture_cb = picture_y + q_data_src->width * q_data_src->height;
>> +     picture_cr = picture_cb + q_data_src->width / 2 *
>> +                     q_data_src->height / 2;
>> +
>> +     coda_write(dev, picture_y, CODA_CMD_ENC_PIC_SRC_ADDR_Y);
>> +     coda_write(dev, picture_cb, CODA_CMD_ENC_PIC_SRC_ADDR_CB);
>> +     coda_write(dev, picture_cr, CODA_CMD_ENC_PIC_SRC_ADDR_CR);
>> +     coda_write(dev, force_ipicture << 1 & 0x2,
>> +                CODA_CMD_ENC_PIC_OPTION);
>> +
>> +     coda_write(dev, pic_stream_buffer_addr, CODA_CMD_ENC_PIC_BB_START);
>> +     coda_write(dev, pic_stream_buffer_size / 1024,
>> +                CODA_CMD_ENC_PIC_BB_SIZE);
>
> If I set CODA_REG_BIT_WR_PTR to pic_stream_buffer_addr here, and
> pic_stream_buffer_addr points to the second destination buffer, after
> the PIC_RUN command finishes, CODA_REG_BIT_WR_PTR again points into the
> first destination buffer, as set in start_streaming().

In i.MX27 I don't even need to set CODA_REG_BIT_WR_PTR here to get the
expected result.

>> +     coda_command_async(ctx, CODA_COMMAND_PIC_RUN);
>> +}
>> +
>> +static int coda_job_ready(void *m2m_priv)
>> +{
>> +     struct coda_ctx *ctx = m2m_priv;
>> +
>> +     /*
>> +      * For both 'P' and 'key' frame cases 1 picture
>> +      * and 1 frame are needed.
>> +      */
>> +     if (!(v4l2_m2m_num_src_bufs_ready(ctx->m2m_ctx) >= 1) ||
>> +             !(v4l2_m2m_num_dst_bufs_ready(ctx->m2m_ctx) >= 1)) {
>> +             v4l2_dbg(1, coda_debug, &ctx->dev->v4l2_dev,
>> +                      "not ready: not enough video buffers.\n");
>> +             return 0;
>> +     }
>> +
>> +     /* For P frames a reference picture is needed too */
>> +     if ((ctx->gopcounter != (ctx->params.gop_size - 1)) &&
>> +        (!ctx->reference)) {
>> +             v4l2_dbg(1, coda_debug, &ctx->dev->v4l2_dev,
>> +                      "not ready: reference picture not available.\n");
>> +             return 0;
>> +     }
>> +
>> +     if (coda_isbusy(ctx->dev)) {
>> +             v4l2_dbg(1, coda_debug, &ctx->dev->v4l2_dev,
>> +                      "not ready: coda is still busy.\n");
>> +             return 0;
>> +     }
>> +
>> +     v4l2_dbg(1, coda_debug, &ctx->dev->v4l2_dev,
>> +                     "job ready\n");
>> +     return 1;
>> +}
> [...]
>> +static int coda_start_streaming(struct vb2_queue *q, unsigned int count)
>> +{
>> +     struct coda_ctx *ctx = vb2_get_drv_priv(q);
>> +     struct coda_dev *dev = ctx->dev;
>> +     u32 bitstream_buf, bitstream_size;
>> +
>> +     if (count < 1)
>> +             return -EINVAL;
>> +
>> +     if (q->type == V4L2_BUF_TYPE_VIDEO_OUTPUT)
>> +             ctx->rawstreamon = 1;
>> +     else
>> +             ctx->compstreamon = 1;
>> +
>> +     if (ctx->rawstreamon & ctx->compstreamon) {
>> +             struct coda_q_data *q_data_src, *q_data_dst;
>> +             u32 dst_fourcc;
>> +             struct vb2_buffer *buf;
>> +             struct vb2_queue *src_vq;
>> +             u32 value;
>> +             int i = 0;
>> +
>> +             ctx->gopcounter = ctx->params.gop_size - 1;
>> +
>> +             q_data_src = get_q_data(ctx, V4L2_BUF_TYPE_VIDEO_OUTPUT);
>> +             buf = v4l2_m2m_next_dst_buf(ctx->m2m_ctx);
>> +             bitstream_buf = vb2_dma_contig_plane_dma_addr(buf, 0);
>> +             q_data_dst = get_q_data(ctx, V4L2_BUF_TYPE_VIDEO_CAPTURE);
>> +             bitstream_size = q_data_dst->sizeimage;
>> +             dst_fourcc = q_data_dst->fmt->fourcc;
>> +
>> +             /* Find out whether coda must encode or decode */
>> +             if (q_data_src->fmt->type == CODA_FMT_RAW &&
>> +                 q_data_dst->fmt->type == CODA_FMT_ENC) {
>> +                     ctx->inst_type = CODA_INST_ENCODER;
>> +             } else if (q_data_src->fmt->type == CODA_FMT_ENC &&
>> +                        q_data_dst->fmt->type == CODA_FMT_RAW) {
>> +                     ctx->inst_type = CODA_INST_DECODER;
>> +                     v4l2_err(&ctx->dev->v4l2_dev, "decoding not supported.\n");
>> +                     return -EINVAL;
>> +             } else {
>> +                     v4l2_err(&ctx->dev->v4l2_dev, "couldn't tell instance type.\n");
>> +                     return -EINVAL;
>> +             }
>> +
>> +             if (!coda_is_initialized(dev)) {
>> +                     v4l2_err(&ctx->dev->v4l2_dev, "coda is not initialized.\n");
>> +                     return -EFAULT;
>> +             }
>> +             coda_write(dev, ctx->parabuf.paddr, CODA_REG_BIT_PARA_BUF_ADDR);
>> +             coda_write(dev, bitstream_buf, CODA_REG_BIT_RD_PTR(ctx->idx));
>> +             coda_write(dev, bitstream_buf, CODA_REG_BIT_WR_PTR(ctx->idx));
>
> Here CODA_REG_BIT_WR_PTR is set to the beginning of bitstream_buf, which
> is the first destination buffer.

Right.

>> +             switch (dev->devtype->product) {
>> +             case CODA_DX6:
>> +                     /*
>> +                      * We should use (CODADX6_STREAM_BUF_PIC_RESET |
>> +                      * CODADX6_STREAM_BUF_PIC_FLUSH) but it doesn't work
>> +                      * with firmware 2.2.5.
>> +                      */
>> +                     coda_write(dev, (3 << 3), CODA_REG_BIT_STREAM_CTRL);
>> +                     break;
>> +             default:
>> +                     coda_write(dev, CODA7_STREAM_BUF_PIC_RESET |
>> +                             CODA7_STREAM_BUF_PIC_FLUSH, CODA_REG_BIT_STREAM_CTRL);
>> +             }
>> +
>> +             /* Configure the coda */
>> +             coda_write(dev, 0xffff4c00, CODA_REG_BIT_SEARCH_RAM_BASE_ADDR);
>> +
>> +             /* Could set rotation here if needed */
[snip]
>> +enum coda_platform {
>> +     CODA_IMX27,
>> +};
>> +
>> +static struct coda_devtype coda_devdata[] = {
>> +     [CODA_IMX27] = {
>> +             .firmware    = "v4l-codadx6-imx27.bin",
>> +             .product     = CODA_DX6,
>> +             .formats     = codadx6_formats,
>> +             .num_formats = ARRAY_SIZE(codadx6_formats),
>> +     },
>> +};
>> +
>> +static struct platform_device_id coda_platform_ids[] = {
>> +     { .name = "coda-imx27", .driver_data = CODA_IMX27 },
>> +     { /* sentinel */ }
>> +};
>> +MODULE_DEVICE_TABLE(platform, coda_platform_ids);
>> +
>> +#ifdef CONFIG_OF
>> +static const struct of_device_id coda_dt_ids[] = {
>> +     { .compatible = "fsl,imx27-vpu", .data = &coda_platform_ids[CODA_IMX27] },
>> +     { /* sentinel */ }
>> +};
>> +MODULE_DEVICE_TABLE(platform, coda_dt_ids);
>
> Should be
> MODULE_DEVICE_TABLE(of, coda_dt_ids);
>
> I completely missed this before, but today it was pointed out to me that
> module autoloading doesn't work.

OK, let me fix it in a new patch.

> [...]
>> +static int coda_remove(struct platform_device *pdev)
>> +{
>> +     struct coda_dev *dev = platform_get_drvdata(pdev);
>> +     unsigned int bufsize;
>> +
>> +     switch (dev->devtype->product) {
>> +     case CODA_DX6:
>> +             bufsize = CODADX6_WORK_BUF_SIZE;
>> +             break;
>> +     default:
>> +             bufsize = CODA7_WORK_BUF_SIZE;
>> +     }
>> +
>> +     video_unregister_device(dev->vfd);
>> +     v4l2_m2m_release(dev->m2m_dev);
>> +     vb2_dma_contig_cleanup_ctx(dev->alloc_ctx);
>> +     video_device_release(dev->vfd);
>
> -+       video_device_release(dev->vfd);
> ++       v4l2_device_unregister(&dev->v4l2_dev);
>
> video_device_release is already called by kobj refcounting.
> With this change I can cleanly rmmod/insmod the coda.ko module.
>

Fine, let me fix it.

>> +     if (dev->codebuf.vaddr)
>> +             dma_free_coherent(&pdev->dev, dev->codebuf_size,
>> +                               &dev->codebuf.vaddr, dev->codebuf.paddr);
>> +     dma_free_coherent(&pdev->dev, bufsize, &dev->workbuf.vaddr,
>> +                       dev->workbuf.paddr);
>> +     return 0;
>> +}
>> +
>> +static struct platform_driver coda_driver = {
>> +     .probe  = coda_probe,
>> +     .remove = __devexit_p(coda_remove),
>> +     .driver = {
>> +             .name   = CODA_NAME,
>> +             .owner  = THIS_MODULE,
>> +             .of_match_table = of_match_ptr(coda_dt_ids),
>> +     },
>> +     .id_table = coda_platform_ids,
>> +};
>> +
>> +module_platform_driver(coda_driver);
>> +
>> +MODULE_LICENSE("GPL");
>> +MODULE_AUTHOR("Javier Martin <javier.martin@vista-silicon.com>");
>> +MODULE_DESCRIPTION("Coda multi-standard codec V4L2 driver");
> [...]
>
> regards
> Philipp


I will send a new v4 with the 'platform' and 'bytesused' issues fixed.
Regarding your i.MX53 problems I suppose they should be addressed
conditionally in a patch on top of this one where i.MX53 support is
added too.
What do you think?

Regards.
-- 
Javier Martin
Vista Silicon S.L.
CDTUC - FASE C - Oficina S-345
Avda de los Castros s/n
39005- Santander. Cantabria. Spain
+34 942 25 32 60
www.vista-silicon.com

  reply	other threads:[~2012-07-18  7:12 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-07-12  7:11 [PATCH v3] media: coda: Add driver for Coda video codec Javier Martin
2012-07-16 17:21 ` Philipp Zabel
2012-07-18  7:12   ` javier Martin [this message]
2012-07-18  8:35     ` Philipp Zabel
2012-07-18  9:01       ` javier Martin
2012-07-18  9:22         ` Philipp Zabel
2012-07-18  9:26           ` javier Martin
2012-07-18 12:12             ` Philipp Zabel
2012-07-18 12:19               ` javier Martin

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='CACKLOr3rOPgwMCRdj3ARR+0655Qp=BfEXq0TsB7TU-hO4NSsqg@mail.gmail.com' \
    --to=javier.martin@vista-silicon.com \
    --cc=kyungmin.park@samsung.com \
    --cc=laurent.pinchart@ideasonboard.com \
    --cc=linux-media@vger.kernel.org \
    --cc=mchehab@infradead.org \
    --cc=p.zabel@pengutronix.de \
    --cc=s.hauer@pengutronix.de \
    --cc=s.nawrocki@samsung.com \
    --cc=sakari.ailus@maxwell.research.nokia.com \
    /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 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.