linux-media.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
To: Bingbu Cao <bingbu.cao@linux.intel.com>, bingbu.cao@intel.com
Cc: Jean-Michel Hautbois <jeanmichel.hautbois@ideasonboard.com>,
	Tian Shu Qiu <tian.shu.qiu@intel.com>,
	sakari.ailus@linux.intel.com, linux-media@vger.kernel.org,
	Tomasz Figa <tfiga@google.com>
Subject: Re: [RFC PATCH] media: staging: ipu3-imgu: add the AWB memory layout
Date: Tue, 21 Sep 2021 14:33:59 +0300	[thread overview]
Message-ID: <YUnDJ+fIgyXUj1K3@pendragon.ideasonboard.com> (raw)
In-Reply-To: <491ab306-2b40-3b3e-fc46-d7af2de3db26@ideasonboard.com>

Hello,

On Thu, Sep 09, 2021 at 07:52:46AM +0200, Jean-Michel Hautbois wrote:
> On 09/09/2021 04:19, Bingbu Cao wrote:
> > On 9/8/21 4:17 PM, Laurent Pinchart wrote:
> >> On Wed, Sep 01, 2021 at 12:34:00AM +0300, Laurent Pinchart wrote:
> >>> On Tue, Aug 31, 2021 at 08:51:40PM +0200, Jean-Michel Hautbois wrote:
> >>>> While parsing the RAW AWB metadata, the AWB layout was missing to fully
> >>>> understand which byte corresponds to which feature. Make the field names
> >>>> and usage explicit, as it is used by the userspace applications.
> >>>>
> >>>> Signed-off-by: Jean-Michel Hautbois <jeanmichel.hautbois@ideasonboard.com>
> >>>> ---
> >>>> This structure layout is defined in CrOs:
> >>>> https://chromium.googlesource.com/chromiumos/platform/arc-camera/+/refs/heads/master/hal/intel/include/ia_imaging/awb_public.h
> >>>>
> >>>> There are a few things not really understood right now:
> >>>> - Is sat_ratio a full scale ratio (I can't get more than some values out
> >>>>   of it, is it a ratio of 25%, 50%, 75%, 100% ?)
> >>>> - What are the real minimum and maximum values for the grid size ? From
> >>>>   CrOs it appears to be [16, 80] for width and [16, 60] for height while
> >>>>   in this file it seems to be [16, 160] for width and not really defined
> >>>>   for height AFAICT ?
> >>>> - Same for the block_width_log2 and block_height_log2 which are [3, 7]
> >>>>   in this file and [3, 6] in the awb_public.h header ?
> 
> Do you have any clue for those questions please :-) ?

This is becoming a blocker, it would be really nice if we could have
answers to these questions. The grid size constraints are the most
immediate priority, but understanding the sat_ratio value will be next
veyr shortly.

> >>>>
> >>>>  .../media/ipu3/include/uapi/intel-ipu3.h      | 38 ++++++++++++++-----
> >>>>  1 file changed, 29 insertions(+), 9 deletions(-)
> >>>>
> >>>> diff --git a/drivers/staging/media/ipu3/include/uapi/intel-ipu3.h b/drivers/staging/media/ipu3/include/uapi/intel-ipu3.h
> >>>> index fa3d6ee5adf2..83191aff2ddd 100644
> >>>> --- a/drivers/staging/media/ipu3/include/uapi/intel-ipu3.h
> >>>> +++ b/drivers/staging/media/ipu3/include/uapi/intel-ipu3.h
> >>>> @@ -61,20 +61,40 @@ struct ipu3_uapi_grid_config {
> >>>>  	__u16 y_end;
> >>>>  } __packed;
> >>>>  
> >>>> +/**
> >>>> + * struct ipu3_uapi_awb_raw_buffer - Memory layout for each cell in AWB
> >>>> + *
> >>>> + * @Gr_avg:	Green average for red lines in the cell.
> >>>> + * @R_avg:	Red average in the cell.
> >>>> + * @B_avg:	Blue average in the cell.
> >>>> + * @Gb_avg:	Green average for blue lines in the cell.
> >>>> + * @sat_ratio:  Saturation ratio in the cell.
> >>>> + * @padding0:   Unused byte for padding.
> >>>> + * @padding1:   Unused byte for padding.
> >>>> + * @padding2:   Unused byte for padding.
> >>>> + */
> >>>> +struct ipu3_uapi_awb_raw_buffer {
> >>>> +    unsigned char Gr_avg;
> >>>> +    unsigned char R_avg;
> >>>> +    unsigned char B_avg;
> >>>> +    unsigned char Gb_avg;
> >>>> +    unsigned char sat_ratio;
> >>>> +    unsigned char padding0;
> >>>> +    unsigned char padding1;
> >>>> +    unsigned char padding2;
> > 
> > It is fine for me to define and exposure the awb memory layout in uAPI.
> > 
> > nit: use __u8 here?
> 
> Sure !
> 
> >>>> +} __packed;
> >>>> +
> >>>>  /*
> >>>>   * The grid based data is divided into "slices" called set, each slice of setX
> >>>>   * refers to ipu3_uapi_grid_config width * height_per_slice.
> >>>>   */
> >>>>  #define IPU3_UAPI_AWB_MAX_SETS				60
> >>>> -/* Based on grid size 80 * 60 and cell size 16 x 16 */
> >>>> -#define IPU3_UAPI_AWB_SET_SIZE				1280
> >>>> -#define IPU3_UAPI_AWB_MD_ITEM_SIZE			8
> >>>> -#define IPU3_UAPI_AWB_SPARE_FOR_BUBBLES \
> >>>> -	(IPU3_UAPI_MAX_BUBBLE_SIZE * IPU3_UAPI_MAX_STRIPES * \
> >>>> -	 IPU3_UAPI_AWB_MD_ITEM_SIZE)
> >>>> +#define AWB_PUBLIC_NUM_OF_ITEMS_IN_SET			160
> >>>> +/* Based on max grid height + Spare for bubbles */
> >>>> +#define AWB_PUBLIC_NUM_OF_SETS_IN_BUFFER IPU3_UAPI_AWB_MAX_SETS + \
> >>>> +	(IPU3_UAPI_MAX_BUBBLE_SIZE * IPU3_UAPI_MAX_STRIPES)
> >>>>  #define IPU3_UAPI_AWB_MAX_BUFFER_SIZE \
> >>>> -	(IPU3_UAPI_AWB_MAX_SETS * \
> >>>> -	 (IPU3_UAPI_AWB_SET_SIZE + IPU3_UAPI_AWB_SPARE_FOR_BUBBLES))
> >>>> +        AWB_PUBLIC_NUM_OF_SETS_IN_BUFFER * AWB_PUBLIC_NUM_OF_ITEMS_IN_SET
> > 
> > It's better to update the name of 'IPU3_UAPI_AWB_MAX_BUFFER_SIZE' to align current
> > definition.
> > 
> >>>>  
> >>>>  /**
> >>>>   * struct ipu3_uapi_awb_raw_buffer - AWB raw buffer
> >>>> @@ -83,7 +103,7 @@ struct ipu3_uapi_grid_config {
> >>>>   *		the average values for each color channel.
> >>>>   */
> >>>>  struct ipu3_uapi_awb_raw_buffer {
> >>>> -	__u8 meta_data[IPU3_UAPI_AWB_MAX_BUFFER_SIZE]
> >>>> +	struct ipu3_uapi_awb_raw_buffer meta_data[IPU3_UAPI_AWB_MAX_BUFFER_SIZE]
> >>>>  		__attribute__((aligned(32)));
> >>>>  } __packed;
> >>>>  

-- 
Regards,

Laurent Pinchart

  reply	other threads:[~2021-09-21 11:34 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-08-31 18:51 [RFC PATCH] media: staging: ipu3-imgu: add the AWB memory layout Jean-Michel Hautbois
2021-08-31 21:33 ` Laurent Pinchart
2021-09-08  8:17   ` Laurent Pinchart
2021-09-09  2:19     ` Bingbu Cao
2021-09-09  5:52       ` Jean-Michel Hautbois
2021-09-21 11:33         ` Laurent Pinchart [this message]
2021-09-21 11:29 ` Laurent Pinchart

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=YUnDJ+fIgyXUj1K3@pendragon.ideasonboard.com \
    --to=laurent.pinchart@ideasonboard.com \
    --cc=bingbu.cao@intel.com \
    --cc=bingbu.cao@linux.intel.com \
    --cc=jeanmichel.hautbois@ideasonboard.com \
    --cc=linux-media@vger.kernel.org \
    --cc=sakari.ailus@linux.intel.com \
    --cc=tfiga@google.com \
    --cc=tian.shu.qiu@intel.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).