* Re: [PATCH] Add framescaling support to stk1160
[not found] <1424799211-26488-1-git-send-email-dale.hamel@srvthe.net>
@ 2015-03-10 3:26 ` Ezequiel Garcia
2015-03-11 16:33 ` Ezequiel Garcia
0 siblings, 1 reply; 3+ messages in thread
From: Ezequiel Garcia @ 2015-03-10 3:26 UTC (permalink / raw)
To: Dale Hamel
Cc: Hans Verkuil, sakari.ailus@linux.intel.com,
mchehab@osg.samsung.com, michael@stegemann.it, ezequiel,
linux-media
Dale,
Don't forget to Cc the media mailing list. See below.
On Tue, Feb 24, 2015 at 2:33 PM, Dale Hamel <dale.hamel@srvthe.net> wrote:
> From: Michael Stegemann <michael@stegemann.it>
>
> Implements frame scaling support for stk1160 to support format changes instead of a static frame size.
>
> This is effectively a dumb sampling, and could perhaps benefit from being an averaging instead.
>
> This was a requested "TO DO" for this driver, and allows support for userspace programs like Hyperion and Boblight.
>
> Submitted on behalf of the original author, Michael Stegemann <michael@stegemann.it>
>
> Signed-off-by: Michael Stegemann <michael@stegemann.it>
> Signed-off-by: Dale Hamel <dale.hamel@srvthe.net>
> ---
> drivers/media/usb/stk1160/stk1160-core.c | 1 +
> drivers/media/usb/stk1160/stk1160-reg.h | 26 +++++++
> drivers/media/usb/stk1160/stk1160-v4l.c | 112 +++++++++++++++++++++++-------
> drivers/media/usb/stk1160/stk1160-video.c | 17 ++++-
> drivers/media/usb/stk1160/stk1160.h | 1 +
> 5 files changed, 130 insertions(+), 27 deletions(-)
>
> diff --git a/drivers/media/usb/stk1160/stk1160-core.c b/drivers/media/usb/stk1160/stk1160-core.c
> index 03504dc..1881770 100644
> --- a/drivers/media/usb/stk1160/stk1160-core.c
> +++ b/drivers/media/usb/stk1160/stk1160-core.c
> @@ -418,6 +418,7 @@ static void stk1160_disconnect(struct usb_interface *interface)
> stk1160_ac97_unregister(dev);
>
> stk1160_clear_queue(dev);
> + vb2_queue_release(&dev->vb_vidq);
>
Is this fixing a real bug? Wasn't aware of any leaks in this driver,
but who knows...
> video_unregister_device(&dev->vdev);
> v4l2_device_disconnect(&dev->v4l2_dev);
> diff --git a/drivers/media/usb/stk1160/stk1160-reg.h b/drivers/media/usb/stk1160/stk1160-reg.h
> index 3e49da6..b1fa11d 100644
> --- a/drivers/media/usb/stk1160/stk1160-reg.h
> +++ b/drivers/media/usb/stk1160/stk1160-reg.h
> @@ -33,6 +33,32 @@
> */
> #define STK1160_DCTRL 0x100
>
> +/*
> + * Decimation Control Register:
> + * Byte 104:
> + * Horizontal Decimation Line Unit Count
> + * Byte 105:
> + * Vertical Decimation Line Unit Count
> + * Byte 106:
> + * Bit 0 - Horizontal Decimation Control
> + * 0 Horizontal decimation is disabled.
> + * 1 Horizontal decimation is enabled.
> + * Bit 1 - Decimates Half or More Column
> + * 0 Decimates less than half from original column,
> + * -> send count unit (0x105) before each unit skipped.
> + * 1 Decimates half or more from original column,
> + * -> skip count unit (0x105) before each unit sent.
> + * Bit 2 - Vertical Decimation Control
> + * see Bit 0, only vertical
> + * Bit 3 - Vertical Greater or Equal to Half
> + * see Bit 1, only vertical
> + * Bit 4 - Decimation Unit
> + * 0 Decimation will work with 2 rows or columns per unit.
> + * 1 Decimation will work with 4 rows or columns per unit.
> + */
> +#define STK1160_DMCTRL 0x104
> +
Great, thanks for documenting this fully.
> +
> /* Capture Frame Start Position */
> #define STK116_CFSPO 0x110
> #define STK116_CFSPO_STX_L 0x110
> diff --git a/drivers/media/usb/stk1160/stk1160-v4l.c b/drivers/media/usb/stk1160/stk1160-v4l.c
> index 65a326c..94dafeb 100644
> --- a/drivers/media/usb/stk1160/stk1160-v4l.c
> +++ b/drivers/media/usb/stk1160/stk1160-v4l.c
> @@ -106,6 +106,76 @@ static void stk1160_set_std(struct stk1160 *dev)
>
> }
>
> +static void stk1160_try_fmt(struct stk1160 *dev, struct v4l2_format *f, bool try)
> +{
> + int base_width, base_height;
> +
> + if (dev->norm & V4L2_STD_525_60){
> + base_width = 720;
> + base_height = 480;
> + } else {
> + base_width = 720;
> + base_height = 576;
> + }
> +
> + if (f->fmt.pix.width <= (base_width / 3)
> + || f->fmt.pix.height <= (base_height / 3)){
> + f->fmt.pix.width = base_width / 3;
> + f->fmt.pix.height = base_height / 3;
> + if (!try){
> + dev->decimate = 3;
> + }
> + } else if ((f->fmt.pix.width >= base_width >> 1
> + && f->fmt.pix.width < base_width)
> + ||((f->fmt.pix.height >= base_height >> 1
> + && f->fmt.pix.height < base_height))){
> + f->fmt.pix.width = base_width >> 1;
> + f->fmt.pix.height = base_height >> 1;
> + if (!try){
> + dev->decimate = 2;
> + }
> + } else {
> + f->fmt.pix.width = base_width;
> + f->fmt.pix.height = base_height;
> + if (!try){
> + dev->decimate = 0;
> + }
> + }
These if/else if/else block looks really ugly.
> +}
> +
> +static void stk1160_set_fmt(struct stk1160 *dev)
> +{
> + if (dev->norm & V4L2_STD_525_60){
> + dev->width = 720;
> + dev->height = 480;
> + } else {
> + dev->width = 720;
> + dev->height = 576;
> + }
> +
> + switch (dev->decimate){
> + case 0:
> + stk1160_write_reg(dev, STK1160_DMCTRL, 0x00);
> + stk1160_write_reg(dev, STK1160_DMCTRL+1, 0x00);
> + stk1160_write_reg(dev, STK1160_DMCTRL+2, 0x00);
> + break;
> + case 2:
> + stk1160_write_reg(dev, STK1160_DMCTRL, 0x01);
> + stk1160_write_reg(dev, STK1160_DMCTRL+1, 0x01);
> + stk1160_write_reg(dev, STK1160_DMCTRL+2, 0x1f);
> + dev->width = dev->width >> 1;
> + dev->height = dev->height >> 1;
> + break;
> + case 3:
> + stk1160_write_reg(dev, STK1160_DMCTRL, 0x02);
> + stk1160_write_reg(dev, STK1160_DMCTRL+1, 0x02);
> + stk1160_write_reg(dev, STK1160_DMCTRL+2, 0x1f);
> + dev->width = dev->width / 3;
> + dev->height = dev->height / 3;
Seems there's room for simplification here.
> + break;
> + }
> +}
> +
> /*
> * Set a new alternate setting.
> * Returns true is dev->max_pkt_size has changed, false otherwise.
> @@ -321,17 +391,10 @@ static int vidioc_try_fmt_vid_cap(struct file *file, void *priv,
> {
> struct stk1160 *dev = video_drvdata(file);
>
> - /*
> - * User can't choose size at his own will,
> - * so we just return him the current size chosen
> - * at standard selection.
> - * TODO: Implement frame scaling?
> - */
> + stk1160_try_fmt(dev, f, true);
>
> - f->fmt.pix.pixelformat = dev->fmt->fourcc;
> - f->fmt.pix.width = dev->width;
> - f->fmt.pix.height = dev->height;
> f->fmt.pix.field = V4L2_FIELD_INTERLACED;
> + f->fmt.pix.pixelformat = dev->fmt->fourcc;
This is just churn. You are just moving this line.
> f->fmt.pix.bytesperline = dev->width * 2;
> f->fmt.pix.sizeimage = dev->height * f->fmt.pix.bytesperline;
> f->fmt.pix.colorspace = V4L2_COLORSPACE_SMPTE170M;
> @@ -348,9 +411,14 @@ static int vidioc_s_fmt_vid_cap(struct file *file, void *priv,
> if (vb2_is_busy(q))
> return -EBUSY;
>
> - vidioc_try_fmt_vid_cap(file, priv, f);
> + stk1160_try_fmt(dev, f, false);
> + stk1160_set_fmt(dev);
>
> - /* We don't support any format changes */
> + f->fmt.pix.field = V4L2_FIELD_INTERLACED;
> + f->fmt.pix.pixelformat = dev->fmt->fourcc;
> + f->fmt.pix.bytesperline = dev->width * 2;
> + f->fmt.pix.sizeimage = dev->height * f->fmt.pix.bytesperline;
> + f->fmt.pix.colorspace = V4L2_COLORSPACE_SMPTE170M;
>
> return 0;
> }
> @@ -389,23 +457,18 @@ static int vidioc_s_std(struct file *file, void *priv, v4l2_std_id norm)
> dev->norm = norm;
>
> /* This is taken from saa7115 video decoder */
> - if (dev->norm & V4L2_STD_525_60) {
> - dev->width = 720;
> - dev->height = 480;
> - } else if (dev->norm & V4L2_STD_625_50) {
> - dev->width = 720;
> - dev->height = 576;
> + if (dev->norm & V4L2_STD_525_60 || dev->norm & V4L2_STD_625_50) {
> + stk1160_set_std(dev);
> + stk1160_set_fmt(dev);
> +
> + v4l2_device_call_all(&dev->v4l2_dev, 0, video, s_std,
> + dev->norm);
> +
> + return 0;
> } else {
> stk1160_err("invalid standard\n");
> return -EINVAL;
> }
> -
> - stk1160_set_std(dev);
> -
> - v4l2_device_call_all(&dev->v4l2_dev, 0, video, s_std,
> - dev->norm);
> -
> - return 0;
This removal looks unneeded.
> }
>
>
> @@ -671,6 +734,7 @@ int stk1160_video_register(struct stk1160 *dev)
> dev->norm = V4L2_STD_NTSC_M;
> dev->width = 720;
> dev->height = 480;
> + dev->decimate = 0;
>
> /* set default format */
> dev->fmt = &format[0];
> diff --git a/drivers/media/usb/stk1160/stk1160-video.c b/drivers/media/usb/stk1160/stk1160-video.c
> index 39f1aae..5a55028 100644
> --- a/drivers/media/usb/stk1160/stk1160-video.c
> +++ b/drivers/media/usb/stk1160/stk1160-video.c
> @@ -84,7 +84,6 @@ struct stk1160_buffer *stk1160_next_buffer(struct stk1160 *dev)
> if (!list_empty(&dev->avail_bufs)) {
> buf = list_first_entry(&dev->avail_bufs,
> struct stk1160_buffer, list);
> - list_del(&buf->list);
What's this?
> }
> spin_unlock_irqrestore(&dev->buf_lock, flags);
>
> @@ -95,7 +94,7 @@ static inline
> void stk1160_buffer_done(struct stk1160 *dev)
> {
> struct stk1160_buffer *buf = dev->isoc_ctl.buf;
> -
> + unsigned long flags = 0;
You need an empty line between variables and code.
> dev->field_count++;
>
> buf->vb.v4l2_buf.sequence = dev->field_count >> 1;
> @@ -104,7 +103,19 @@ void stk1160_buffer_done(struct stk1160 *dev)
> v4l2_get_timestamp(&buf->vb.v4l2_buf.timestamp);
>
> vb2_set_plane_payload(&buf->vb, 0, buf->bytesused);
> - vb2_buffer_done(&buf->vb, VB2_BUF_STATE_DONE);
> +
> + spin_lock_irqsave(&dev->buf_lock, flags);
Seems like taking the lock here could be correct.
This is a fix, and it's not related to this patch,
so you need to send a separate patch for it.
> + if (buf->bytesused == vb2_plane_size(&buf->vb, 0))
> + vb2_buffer_done(&buf->vb, VB2_BUF_STATE_DONE);
> + else
> + vb2_buffer_done(&buf->vb, VB2_BUF_STATE_ERROR);
> +
> + if (!list_empty(&dev->avail_bufs)) {
> + list_del(&buf->list);
> + }
Hm.. why do you remove the buffer from the list here?
> +
> + spin_unlock_irqrestore(&dev->buf_lock, flags);
> +
>
> dev->isoc_ctl.buf = NULL;
> }
> diff --git a/drivers/media/usb/stk1160/stk1160.h b/drivers/media/usb/stk1160/stk1160.h
> index abdea48..895ed35 100644
> --- a/drivers/media/usb/stk1160/stk1160.h
> +++ b/drivers/media/usb/stk1160/stk1160.h
> @@ -150,6 +150,7 @@ struct stk1160 {
> unsigned int ctl_input; /* selected input */
> v4l2_std_id norm; /* current norm */
> struct stk1160_fmt *fmt; /* selected format */
> + int decimate;
>
> unsigned int field_count; /* not sure ??? */
> enum v4l2_field field; /* also not sure :/ */
> --
> 2.1.1
>
--
Ezequiel
^ permalink raw reply [flat|nested] 3+ messages in thread