Re: [PATCH v2 2/2] media: vimc: Add a control to show test pattern colors' order

[Kieran Bingham]

Hi Kaaira,


On 15/06/2020 15:17, Kaaira Gupta wrote:
> On Mon, Jun 15, 2020 at 12:48:20PM +0100, Kieran Bingham wrote:
>> Hi Kaaira,
>>
>> On 14/06/2020 21:02, Kaaira Gupta wrote:
>>> Add a control in VIMC to show the correct order of the colors for a
>>> given test pattern.
>>> The control can be accessed by using show_colors_order in v4l2-ctl
>>>
>>> Signed-off-by: Kaaira Gupta <kgupta@xxxxxxxxxxxxx>
>>> ---
>>>  drivers/media/test-drivers/vimc/Kconfig       |  2 ++
>>>  drivers/media/test-drivers/vimc/vimc-common.h |  1 +
>>>  drivers/media/test-drivers/vimc/vimc-sensor.c | 34 +++++++++++++++++++
>>>  3 files changed, 37 insertions(+)
>>>
>>> diff --git a/drivers/media/test-drivers/vimc/Kconfig b/drivers/media/test-drivers/vimc/Kconfig
>>> index 4068a67585f9..da4b2ad6e40c 100644
>>> --- a/drivers/media/test-drivers/vimc/Kconfig
>>> +++ b/drivers/media/test-drivers/vimc/Kconfig
>>> @@ -2,6 +2,8 @@
>>>  config VIDEO_VIMC
>>>  	tristate "Virtual Media Controller Driver (VIMC)"
>>>  	depends on VIDEO_DEV && VIDEO_V4L2
>>> +	select FONT_SUPPORT
>>> +	select FONT_8x16
>>>  	select MEDIA_CONTROLLER
>>>  	select VIDEO_V4L2_SUBDEV_API
>>>  	select VIDEOBUF2_VMALLOC
>>> diff --git a/drivers/media/test-drivers/vimc/vimc-common.h b/drivers/media/test-drivers/vimc/vimc-common.h
>>> index ae163dec2459..52376ba6146b 100644
>>> --- a/drivers/media/test-drivers/vimc/vimc-common.h
>>> +++ b/drivers/media/test-drivers/vimc/vimc-common.h
>>> @@ -20,6 +20,7 @@
>>>  #define VIMC_CID_VIMC_CLASS		(0x00f00000 | 1)
>>>  #define VIMC_CID_TEST_PATTERN		(VIMC_CID_VIMC_BASE + 0)
>>>  #define VIMC_CID_MEAN_WIN_SIZE		(VIMC_CID_VIMC_BASE + 1)
>>> +#define VIMC_TEST_PATTERN_ORDER		(VIMC_CID_VIMC_BASE + 2)
>>
>> This should have the prefix CID like the others. I believe that stands
>> for "Control ID"
>>
>> As below, I think it might warrant being a more generic name for the OSD
>> overlay, as it could display more than just the colour sequence.
> 
> Yes, Okay..I think I'll make a separate patch for adding all the text
> that VIMC can show including the color order, and let the 1st patch of
> this series be a separate patch?

This patch can be first and on it's own, but the key point is that once
you define a control name - this becomes the public interface, so it
can't change. (The kernel ABI interfaces should be stable).

You couldn't later decide "oh - now we display more than just the
TEST_PATTERN_ORDER, so we should rename it to VIMC_CID_INFO", (unless
the patches haven't been merged) - so if you know you will add more
information, we should just use a generic name from the start.


> 
> Should the colors' order text be added to VIVID as well?

Potentially yes, I think that would be useful if the code could be shared.

I think your tpg_g_color_order() is already usable by vivid?


>>>  #define VIMC_FRAME_MAX_WIDTH 4096
>>>  #define VIMC_FRAME_MAX_HEIGHT 2160
>>> diff --git a/drivers/media/test-drivers/vimc/vimc-sensor.c b/drivers/media/test-drivers/vimc/vimc-sensor.c
>>> index a2f09ac9a360..da625f6accce 100644
>>> --- a/drivers/media/test-drivers/vimc/vimc-sensor.c
>>> +++ b/drivers/media/test-drivers/vimc/vimc-sensor.c
>>> @@ -5,6 +5,7 @@
>>>   * Copyright (C) 2015-2017 Helen Koike <helen.fornazier@xxxxxxxxx>
>>>   */
>>>  
>>> +#include <linux/font.h>
>>>  #include <linux/v4l2-mediabus.h>
>>>  #include <linux/vmalloc.h>
>>>  #include <media/v4l2-ctrls.h>
>>> @@ -19,6 +20,7 @@ struct vimc_sen_device {
>>>  	struct v4l2_subdev sd;
>>>  	struct tpg_data tpg;
>>>  	u8 *frame;
>>> +	bool show_colors_order;
>>
>> I would say it's the 'sequence' 'show_color_sequence' but it's still a
>> bit long ... but maybe that doesn't really matter.
> 
> If its generic for all the information maybe it should be
> show_info?

show_info sounds good to me.


> 
>>
>>
>>>  	/* The active format */
>>>  	struct v4l2_mbus_framefmt mbus_format;
>>>  	struct v4l2_ctrl_handler hdl;
>>> @@ -185,10 +187,18 @@ static const struct v4l2_subdev_pad_ops vimc_sen_pad_ops = {
>>>  static void *vimc_sen_process_frame(struct vimc_ent_device *ved,
>>>  				    const void *sink_frame)
>>>  {
>>> +	u8 *basep[TPG_MAX_PLANES][2];
>>> +	char *str;
>>>  	struct vimc_sen_device *vsen = container_of(ved, struct vimc_sen_device,
>>>  						    ved);
>>>  
>>>  	tpg_fill_plane_buffer(&vsen->tpg, 0, 0, vsen->frame);
>>> +	if (vsen->show_colors_order) {
>>> +		tpg_calc_text_basep(&vsen->tpg, basep, 0, vsen->frame);
>>> +		str = tpg_g_color_order(&vsen->tpg);
>>> +		tpg_gen_text(&vsen->tpg, basep, 1, 1, str);
>>> +	}
>>> +
>>>  	return vsen->frame;
>>>  }
>>>  
>>> @@ -200,6 +210,14 @@ static int vimc_sen_s_stream(struct v4l2_subdev *sd, int enable)
>>>  	if (enable) {
>>>  		const struct vimc_pix_map *vpix;
>>>  		unsigned int frame_size;
>>> +		const struct font_desc *font = find_font("VGA8x16");
>>> +
>>> +		if (font == NULL) {
>>> +			pr_err("vimc: could not find font\n");
>>> +			return -ENODEV;
>>
>> I wonder if this should instead just disable text rendering instead?
> 
> Shouldn't the user know why he can't see the text?

I think it's probably fine to keep the pr_err() or pr_warn() perhaps -
but instead of return -ENODEV, set show_info = false; ....

I don't think there's a need to stop the stream running if the system
doesn't have a font.

But maybe someone will disagree ... if the info is requested, but can't
be displayed - perhaps it should be a failure :-S


>>
>> It might be reasonable to compile the kernel without the font-library ?
> 
> I have added the support in its Kconfig, isn't that okay?
> 
>>
>>> +		}
>>> +
>>> +		tpg_set_font(font->data);
>>>  
>>>  		/* Calculate the frame size */
>>>  		vpix = vimc_pix_map_by_code(vsen->mbus_format.code);
>>> @@ -269,6 +287,9 @@ static int vimc_sen_s_ctrl(struct v4l2_ctrl *ctrl)
>>>  	case V4L2_CID_SATURATION:
>>>  		tpg_s_saturation(&vsen->tpg, ctrl->val);
>>>  		break;
>>> +	case VIMC_TEST_PATTERN_ORDER:
>>> +		vsen->show_colors_order = ctrl->val;
>>> +		break;
>>>  	default:
>>>  		return -EINVAL;
>>>  	}
>>> @@ -307,6 +328,17 @@ static const struct v4l2_ctrl_config vimc_sen_ctrl_test_pattern = {
>>>  	.qmenu = tpg_pattern_strings,
>>>  };
>>>  
>>> +static const struct v4l2_ctrl_config vimc_sen_ctrl_order = {
>>> +	.ops = &vimc_sen_ctrl_ops,
>>> +	.id = VIMC_TEST_PATTERN_ORDER,
>>
>> Now that you've brought in support for rendering text on the frame, I
>> wonder if more information should be displayed just like VIVID does.
>>
>> In that case, it would probably be better to have a more generic control
>> that enables all of the text OSD with a more generic name.
> 
> Yes, I will change that
> 
>>
>>
>>> +	.name = "Show Colors Order",
>>> +	.type = V4L2_CTRL_TYPE_BOOLEAN,
>>> +	.min = 0,
>>> +	.max = 1,
>>> +	.step = 1,
>>> +	.def = 1,
>>> +};
>>> +
>>>  static struct vimc_ent_device *vimc_sen_add(struct vimc_device *vimc,
>>>  					    const char *vcfg_name)
>>>  {
>>> @@ -323,6 +355,7 @@ static struct vimc_ent_device *vimc_sen_add(struct vimc_device *vimc,
>>>  
>>>  	v4l2_ctrl_new_custom(&vsen->hdl, &vimc_sen_ctrl_class, NULL);
>>>  	v4l2_ctrl_new_custom(&vsen->hdl, &vimc_sen_ctrl_test_pattern, NULL);
>>> +	v4l2_ctrl_new_custom(&vsen->hdl, &vimc_sen_ctrl_order, NULL);
>>>  	v4l2_ctrl_new_std(&vsen->hdl, &vimc_sen_ctrl_ops,
>>>  			  V4L2_CID_VFLIP, 0, 1, 1, 0);
>>>  	v4l2_ctrl_new_std(&vsen->hdl, &vimc_sen_ctrl_ops,
>>> @@ -362,6 +395,7 @@ static struct vimc_ent_device *vimc_sen_add(struct vimc_device *vimc,
>>>  
>>>  	/* Initialize the frame format */
>>>  	vsen->mbus_format = fmt_default;
>>> +	vsen->show_colors_order = vimc_sen_ctrl_order.def;
>>>  
>>>  	return &vsen->ved;
>>>  
>>>
> 
> Thanks for your time :D

Thanks for yours ;-)


> 
>>
>> -- 
>> Regards
>> --
>> Kieran

-- 
Regards
--
Kieran