Re: [PATCH v2 1/3] media: vimc: Support multiple media bus codes for each pixelformat

From: Helen Koike
Date: Mon Apr 20 2020 - 16:50:41 EST

Next message: Wolfram Sang: "Re: [PATCH 1/2] i2c: slave-eeprom: initialize empty eeprom properly"
Previous message: Rob Herring: "Re: [PATCH 05/13] dt-bindings: MIPS: Require SoC compatible string after board string"
In reply to: NÃcolas F. R. A. Prado: "Re: [PATCH v2 1/3] media: vimc: Support multiple media bus codes for each pixelformat"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Hi NÃcolas,

On 4/20/20 5:36 PM, NÃcolas F. R. A. Prado wrote:
> Hi Helen,
>
> thanks for the review.
>
> Some comments below.
>
> On Mon, Mar 30, 2020 at 04:36:45PM -0300, Helen Koike wrote:
>>
>> Hi NÃcolas,
>>
>> thank you for the patch.
>>
>> The series looks good in general, just minor comments below.
>>
>> On 3/26/20 6:47 PM, NÃcolas F. R. A. Prado wrote:
>>> Change vimc_pix_map_list to allow multiple media bus codes to map to the
>>> same pixelformat, making it possible to add media bus codes for which
>>> there are no pixelformat.
>>>
>>> Signed-off-by: NÃcolas F. R. A. Prado <nfraprado@xxxxxxxxxxxxxx>
>>> ---
>>>
>>> Changes in v2:
>>> - Fix vimc_mbus_code_by_index not checking code array bounds
>>> - Fix array formatting
>>> - Rename variables
>>> - Change code array size
>>> - Add comment about vimc_mbus_code_by_index return value
>>>
>>> drivers/media/platform/vimc/vimc-common.c | 70 ++++++++++++++---------
>>> drivers/media/platform/vimc/vimc-common.h | 11 +++-
>>> drivers/media/platform/vimc/vimc-scaler.c | 10 +++-
>>> drivers/media/platform/vimc/vimc-sensor.c | 6 +-
>>> 4 files changed, 65 insertions(+), 32 deletions(-)
>>>
>>> diff --git a/drivers/media/platform/vimc/vimc-common.c b/drivers/media/platform/vimc/vimc-common.c
>>> index c95c17c048f2..119846f3eaa5 100644
>>> --- a/drivers/media/platform/vimc/vimc-common.c
>>> +++ b/drivers/media/platform/vimc/vimc-common.c
>>> @@ -19,19 +19,19 @@ static const struct vimc_pix_map vimc_pix_map_list[] = {
>>>
>>> /* RGB formats */
>>> {
>>> - .code = MEDIA_BUS_FMT_BGR888_1X24,
>>> + .code = { MEDIA_BUS_FMT_BGR888_1X24 },
>>> .pixelformat = V4L2_PIX_FMT_BGR24,
>>> .bpp = 3,
>>> .bayer = false,
>>> },
>>> {
>>> - .code = MEDIA_BUS_FMT_RGB888_1X24,
>>> + .code = { MEDIA_BUS_FMT_RGB888_1X24 },
>>> .pixelformat = V4L2_PIX_FMT_RGB24,
>>> .bpp = 3,
>>> .bayer = false,
>>> },
>>> {
>>> - .code = MEDIA_BUS_FMT_ARGB8888_1X32,
>>> + .code = { MEDIA_BUS_FMT_ARGB8888_1X32 },
>>> .pixelformat = V4L2_PIX_FMT_ARGB32,
>>> .bpp = 4,
>>> .bayer = false,
>>> @@ -39,49 +39,49 @@ static const struct vimc_pix_map vimc_pix_map_list[] = {
>>>
>>> /* Bayer formats */
>>> {
>>> - .code = MEDIA_BUS_FMT_SBGGR8_1X8,
>>> + .code = { MEDIA_BUS_FMT_SBGGR8_1X8 },
>>> .pixelformat = V4L2_PIX_FMT_SBGGR8,
>>> .bpp = 1,
>>> .bayer = true,
>>> },
>>> {
>>> - .code = MEDIA_BUS_FMT_SGBRG8_1X8,
>>> + .code = { MEDIA_BUS_FMT_SGBRG8_1X8 },
>>> .pixelformat = V4L2_PIX_FMT_SGBRG8,
>>> .bpp = 1,
>>> .bayer = true,
>>> },
>>> {
>>> - .code = MEDIA_BUS_FMT_SGRBG8_1X8,
>>> + .code = { MEDIA_BUS_FMT_SGRBG8_1X8 },
>>> .pixelformat = V4L2_PIX_FMT_SGRBG8,
>>> .bpp = 1,
>>> .bayer = true,
>>> },
>>> {
>>> - .code = MEDIA_BUS_FMT_SRGGB8_1X8,
>>> + .code = { MEDIA_BUS_FMT_SRGGB8_1X8 },
>>> .pixelformat = V4L2_PIX_FMT_SRGGB8,
>>> .bpp = 1,
>>> .bayer = true,
>>> },
>>> {
>>> - .code = MEDIA_BUS_FMT_SBGGR10_1X10,
>>> + .code = { MEDIA_BUS_FMT_SBGGR10_1X10 },
>>> .pixelformat = V4L2_PIX_FMT_SBGGR10,
>>> .bpp = 2,
>>> .bayer = true,
>>> },
>>> {
>>> - .code = MEDIA_BUS_FMT_SGBRG10_1X10,
>>> + .code = { MEDIA_BUS_FMT_SGBRG10_1X10 },
>>> .pixelformat = V4L2_PIX_FMT_SGBRG10,
>>> .bpp = 2,
>>> .bayer = true,
>>> },
>>> {
>>> - .code = MEDIA_BUS_FMT_SGRBG10_1X10,
>>> + .code = { MEDIA_BUS_FMT_SGRBG10_1X10 },
>>> .pixelformat = V4L2_PIX_FMT_SGRBG10,
>>> .bpp = 2,
>>> .bayer = true,
>>> },
>>> {
>>> - .code = MEDIA_BUS_FMT_SRGGB10_1X10,
>>> + .code = { MEDIA_BUS_FMT_SRGGB10_1X10 },
>>> .pixelformat = V4L2_PIX_FMT_SRGGB10,
>>> .bpp = 2,
>>> .bayer = true,
>>> @@ -89,25 +89,25 @@ static const struct vimc_pix_map vimc_pix_map_list[] = {
>>>
>>> /* 10bit raw bayer a-law compressed to 8 bits */
>>> {
>>> - .code = MEDIA_BUS_FMT_SBGGR10_ALAW8_1X8,
>>> + .code = { MEDIA_BUS_FMT_SBGGR10_ALAW8_1X8 },
>>> .pixelformat = V4L2_PIX_FMT_SBGGR10ALAW8,
>>> .bpp = 1,
>>> .bayer = true,
>>> },
>>> {
>>> - .code = MEDIA_BUS_FMT_SGBRG10_ALAW8_1X8,
>>> + .code = { MEDIA_BUS_FMT_SGBRG10_ALAW8_1X8 },
>>> .pixelformat = V4L2_PIX_FMT_SGBRG10ALAW8,
>>> .bpp = 1,
>>> .bayer = true,
>>> },
>>> {
>>> - .code = MEDIA_BUS_FMT_SGRBG10_ALAW8_1X8,
>>> + .code = { MEDIA_BUS_FMT_SGRBG10_ALAW8_1X8 },
>>> .pixelformat = V4L2_PIX_FMT_SGRBG10ALAW8,
>>> .bpp = 1,
>>> .bayer = true,
>>> },
>>> {
>>> - .code = MEDIA_BUS_FMT_SRGGB10_ALAW8_1X8,
>>> + .code = { MEDIA_BUS_FMT_SRGGB10_ALAW8_1X8 },
>>> .pixelformat = V4L2_PIX_FMT_SRGGB10ALAW8,
>>> .bpp = 1,
>>> .bayer = true,
>>> @@ -115,49 +115,49 @@ static const struct vimc_pix_map vimc_pix_map_list[] = {
>>>
>>> /* 10bit raw bayer DPCM compressed to 8 bits */
>>> {
>>> - .code = MEDIA_BUS_FMT_SBGGR10_DPCM8_1X8,
>>> + .code = { MEDIA_BUS_FMT_SBGGR10_DPCM8_1X8 },
>>> .pixelformat = V4L2_PIX_FMT_SBGGR10DPCM8,
>>> .bpp = 1,
>>> .bayer = true,
>>> },
>>> {
>>> - .code = MEDIA_BUS_FMT_SGBRG10_DPCM8_1X8,
>>> + .code = { MEDIA_BUS_FMT_SGBRG10_DPCM8_1X8 },
>>> .pixelformat = V4L2_PIX_FMT_SGBRG10DPCM8,
>>> .bpp = 1,
>>> .bayer = true,
>>> },
>>> {
>>> - .code = MEDIA_BUS_FMT_SGRBG10_DPCM8_1X8,
>>> + .code = { MEDIA_BUS_FMT_SGRBG10_DPCM8_1X8 },
>>> .pixelformat = V4L2_PIX_FMT_SGRBG10DPCM8,
>>> .bpp = 1,
>>> .bayer = true,
>>> },
>>> {
>>> - .code = MEDIA_BUS_FMT_SRGGB10_DPCM8_1X8,
>>> + .code = { MEDIA_BUS_FMT_SRGGB10_DPCM8_1X8 },
>>> .pixelformat = V4L2_PIX_FMT_SRGGB10DPCM8,
>>> .bpp = 1,
>>> .bayer = true,
>>> },
>>> {
>>> - .code = MEDIA_BUS_FMT_SBGGR12_1X12,
>>> + .code = { MEDIA_BUS_FMT_SBGGR12_1X12 },
>>> .pixelformat = V4L2_PIX_FMT_SBGGR12,
>>> .bpp = 2,
>>> .bayer = true,
>>> },
>>> {
>>> - .code = MEDIA_BUS_FMT_SGBRG12_1X12,
>>> + .code = { MEDIA_BUS_FMT_SGBRG12_1X12 },
>>> .pixelformat = V4L2_PIX_FMT_SGBRG12,
>>> .bpp = 2,
>>> .bayer = true,
>>> },
>>> {
>>> - .code = MEDIA_BUS_FMT_SGRBG12_1X12,
>>> + .code = { MEDIA_BUS_FMT_SGRBG12_1X12 },
>>> .pixelformat = V4L2_PIX_FMT_SGRBG12,
>>> .bpp = 2,
>>> .bayer = true,
>>> },
>>> {
>>> - .code = MEDIA_BUS_FMT_SRGGB12_1X12,
>>> + .code = { MEDIA_BUS_FMT_SRGGB12_1X12 },
>>> .pixelformat = V4L2_PIX_FMT_SRGGB12,
>>> .bpp = 2,
>>> .bayer = true,
>>> @@ -182,13 +182,31 @@ const struct vimc_pix_map *vimc_pix_map_by_index(unsigned int i)
>>> return &vimc_pix_map_list[i];
>>> }
>>>
>>> +const u32 vimc_mbus_code_by_index(unsigned int index)
>>> +{
>>> + unsigned int i, j;
>>> +
>>> + for (i = 0; i < ARRAY_SIZE(vimc_pix_map_list); i++) {
>>> + for (j = 0; j < ARRAY_SIZE(vimc_pix_map_list[i].code); j++) {
>>> + if (vimc_pix_map_list[i].code[j]) {
>>
>> Can this be false?
>
> Actually it can, but after you asked I realized this code could be way clearer.
>
> When vimc_pix_map_list[i].code[j] is 0, it means that this is an unused value of
> the array, so we should skip to the next pixelformat.
> I think writing it this way instead would be better, what do you think?
>
> for (i = 0; i < ARRAY_SIZE(vimc_pix_map_list); i++) {
> for (j = 0; j < ARRAY_SIZE(vimc_pix_map_list[i].code); j++) {
> if (!vimc_pix_map_list[i].code[j])
> break;
>
> if (!index)
> return vimc_pix_map_list[i].code[j];
> index--;
>

Looks better, I prefer reducing indentation.

>>
>>> + if (!index)
>>> + return vimc_pix_map_list[i].code[j];
>>> + index--;
>>> + }
>>> + }
>>> + }
>>> + return 0;
>>> +}
>>> +
>>> const struct vimc_pix_map *vimc_pix_map_by_code(u32 code)
>>> {
>>> - unsigned int i;
>>> + unsigned int i, j;
>>>
>>> for (i = 0; i < ARRAY_SIZE(vimc_pix_map_list); i++) {
>>> - if (vimc_pix_map_list[i].code == code)
>>> - return &vimc_pix_map_list[i];
>>> + for (j = 0; j < ARRAY_SIZE(vimc_pix_map_list[i].code); j++) {
>>> + if (vimc_pix_map_list[i].code[j] == code)
>>> + return &vimc_pix_map_list[i];
>>> + }
>>> }
>>> return NULL;
>>> }
>>> diff --git a/drivers/media/platform/vimc/vimc-common.h b/drivers/media/platform/vimc/vimc-common.h
>>> index 616d5a6b0754..585441694c86 100644
>>> --- a/drivers/media/platform/vimc/vimc-common.h
>>> +++ b/drivers/media/platform/vimc/vimc-common.h
>>> @@ -69,7 +69,7 @@ do { \
>>> * V4L2_PIX_FMT_* fourcc pixelformat and its bytes per pixel (bpp)
>>> */
>>> struct vimc_pix_map {
>>> - unsigned int code;
>>> + unsigned int code[1];
>>> unsigned int bpp;
>>> u32 pixelformat;
>>> bool bayer;
>>> @@ -172,6 +172,15 @@ void vimc_sen_release(struct vimc_ent_device *ved);
>>> */
>>> const struct vimc_pix_map *vimc_pix_map_by_index(unsigned int i);
>>>
>>> +/**
>>> + * vimc_mbus_code_by_index - get mbus code by its index
>>> + *
>>> + * @index: index of the mbus code in vimc_pix_map_list
>>> + *
>>> + * Returns 0 if no mbus code is found for the given index.
>>> + */
>>> +const u32 vimc_mbus_code_by_index(unsigned int index);
>>> +
>>> /**
>>> * vimc_pix_map_by_code - get vimc_pix_map struct by media bus code
>>> *
>>> diff --git a/drivers/media/platform/vimc/vimc-scaler.c b/drivers/media/platform/vimc/vimc-scaler.c
>>> index 7521439747c5..6bac1fa65a6f 100644
>>> --- a/drivers/media/platform/vimc/vimc-scaler.c
>>> +++ b/drivers/media/platform/vimc/vimc-scaler.c
>>> @@ -110,13 +110,19 @@ static int vimc_sca_enum_mbus_code(struct v4l2_subdev *sd,
>>> struct v4l2_subdev_pad_config *cfg,
>>> struct v4l2_subdev_mbus_code_enum *code)
>>> {
>>> - const struct vimc_pix_map *vpix = vimc_pix_map_by_index(code->index);
>>> + const u32 mbus_code = vimc_mbus_code_by_index(code->index);
>>> + const struct vimc_pix_map *vpix;
>>> +
>>> + if (!mbus_code)
>>> + return -EINVAL;
>>> +
>>> + vpix = vimc_pix_map_by_code(mbus_code);
>>>
>>> /* We don't support bayer format */
>>> if (!vpix || vpix->bayer)
>>> return -EINVAL;
>>>
>>> - code->code = vpix->code;
>>> + code->code = mbus_code;
>>
>> no need to change this.
>
> This change is actually needed, because after this patch, the code property of
> vimc_pix_map_list is an array, so there isn't a 1 to 1 relation between mbus
> code and pixmap format anymore.
> Since we already got the mbus code by index through
> vimc_mbus_code_by_index(code->index), we just use it.

Make sense, thank you for your explanation.

Regards,
Helen

>
>>
>>>
>>> return 0;
>>> }
>>> diff --git a/drivers/media/platform/vimc/vimc-sensor.c b/drivers/media/platform/vimc/vimc-sensor.c
>>> index 92daee58209e..b8bd430809c1 100644
>>> --- a/drivers/media/platform/vimc/vimc-sensor.c
>>> +++ b/drivers/media/platform/vimc/vimc-sensor.c
>>> @@ -52,12 +52,12 @@ static int vimc_sen_enum_mbus_code(struct v4l2_subdev *sd,
>>> struct v4l2_subdev_pad_config *cfg,
>>> struct v4l2_subdev_mbus_code_enum *code)
>>> {
>>> - const struct vimc_pix_map *vpix = vimc_pix_map_by_index(code->index);
>>> + const u32 mbus_code = vimc_mbus_code_by_index(code->index);
>>>
>>> - if (!vpix)
>>> + if (!mbus_code)
>>> return -EINVAL;
>>>
>>> - code->code = vpix->code;
>>> + code->code = mbus_code;
>>>
>>> return 0;
>>> }
>>>
>>
>> With these changes
>>
>> Acked-by: Helen Koike <helen.koike@xxxxxxxxxxxxx>
>>
>> Regards,
>> Helen
>
> Thank you,
> NÃcolas
>

Next message: Wolfram Sang: "Re: [PATCH 1/2] i2c: slave-eeprom: initialize empty eeprom properly"
Previous message: Rob Herring: "Re: [PATCH 05/13] dt-bindings: MIPS: Require SoC compatible string after board string"
In reply to: NÃcolas F. R. A. Prado: "Re: [PATCH v2 1/3] media: vimc: Support multiple media bus codes for each pixelformat"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]