From: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
To: "Barnabás Pőcze" <barnabas.pocze+renesas@ideasonboard.com>
Cc: "Niklas Söderlund" <niklas.soderlund@ragnatech.se>,
"Mauro Carvalho Chehab" <mchehab@kernel.org>,
"Geert Uytterhoeven" <geert+renesas@glider.be>,
"Magnus Damm" <magnus.damm@gmail.com>,
"Jacopo Mondi" <jacopo.mondi+renesas@ideasonboard.com>,
"Sakari Ailus" <sakari.ailus@linux.intel.com>,
linux-media@vger.kernel.org, linux-renesas-soc@vger.kernel.org,
linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2] media: rcar-isp: ispcore: Fix inconsistent step sizes
Date: Sun, 4 Oct 2026 11:14:37 +0200 [thread overview]
Message-ID: <asIXMIVjbusYLbaR@zed> (raw)
In-Reply-To: <20261001085130.84565-1-barnabas.pocze+renesas@ideasonboard.com>
Hi Barnabás
On Thu, Oct 01, 2026 at 10:51:30AM +0200, Barnabás Pőcze wrote:
> `risp_io_{input,capture}_enum_framesizes()` sets the horizontal and vertical
> step size to 2. However, `risp_io_{input,capture}_try_format()` passes 2
> to `v4l_bound_align_image()`, which corresponds to a step size of 2^2 = 4.
>
> So move these constants into macros to avoid the repetition, and with that,
> use 4 as the step size everywhere.
Why 4 ?
I can't find any such limit in documentation.
For input I don't find alignment constraints nor in CS or VSPX
documentation.
For output I only see a requirement that the stride is 256 bits
aligned, something that is enforced by risp_io_capture_try_format()
already.
I see Niklas has reviewed the patch, so he might have found somewhere
in the long documentation where the alignment for both input and
capture nodes is described.
>
> Fixes: 2151350f60d1 ("media: rcar-isp: Add support for ISPCORE")
> Signed-off-by: Barnabás Pőcze <barnabas.pocze+renesas@ideasonboard.com>
> Reviewed-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
> ---
> changes in v2:
> * fix typo
>
> v1: https://lore.kernel.org/linux-media/20260929133020.342677-2-barnabas.pocze+renesas@ideasonboard.com
> ---
> .../media/platform/renesas/rcar-isp/core-io.c | 40 +++++++++++--------
> 1 file changed, 24 insertions(+), 16 deletions(-)
>
> diff --git a/drivers/media/platform/renesas/rcar-isp/core-io.c b/drivers/media/platform/renesas/rcar-isp/core-io.c
> index 820af506f896c..b60b91f43d42d 100644
> --- a/drivers/media/platform/renesas/rcar-isp/core-io.c
> +++ b/drivers/media/platform/renesas/rcar-isp/core-io.c
> @@ -15,6 +15,12 @@
>
> #include "risp-core.h"
>
> +#define RISP_MIN_WIDTH 128
> +#define RISP_MAX_WIDTH 5120
> +#define RISP_MIN_HEIGHT 128
> +#define RISP_MAX_HEIGHT 4096
> +#define RISP_SIZE_ALIGNMENT 2 /* 2^2 = 4 */
The fact that 2 means 2^2 is only because of how
v4l_bound_align_image() is implemented
> +
> #define risp_io_err(d, fmt, arg...) dev_err((d)->core->dev, fmt, ##arg)
>
> static struct risp_buffer *risp_io_vb2buf(struct vb2_v4l2_buffer *vb)
> @@ -330,8 +336,9 @@ static void risp_io_input_try_format(struct rcar_isp_core_io *io,
> {
> unsigned int bpp = 0;
>
> - v4l_bound_align_image(&pix->width, 128, 5120, 2,
> - &pix->height, 128, 4096, 2, 0);
> + v4l_bound_align_image(&pix->width, RISP_MIN_WIDTH, RISP_MAX_WIDTH, RISP_SIZE_ALIGNMENT,
> + &pix->height, RISP_MIN_HEIGHT, RISP_MAX_HEIGHT, RISP_SIZE_ALIGNMENT,
Should we maybe use 1 here, unless it is documented that 4 is a
requirement ?
My suspicion is that we originally meant '2' everywhere, but
v4l_bound_align_image() is weird and if you pass '2' in it means
'2^2' and this went unnoticed.
> + 0);
>
> for (unsigned int i = 0; i < ARRAY_SIZE(risp_io_input_formats); i++) {
> if (risp_io_input_formats[i].fourcc == pix->pixelformat) {
> @@ -423,13 +430,13 @@ static int risp_io_input_enum_framesizes(struct file *file, void *fh,
>
> fsize->type = V4L2_FRMSIZE_TYPE_STEPWISE;
>
> - fsize->stepwise.min_width = 128;
> - fsize->stepwise.max_width = 5120;
> - fsize->stepwise.step_width = 2;
> + fsize->stepwise.min_width = RISP_MIN_WIDTH;
> + fsize->stepwise.max_width = RISP_MAX_WIDTH;
> + fsize->stepwise.step_width = 1u << RISP_SIZE_ALIGNMENT;
>
> - fsize->stepwise.min_height = 128;
> - fsize->stepwise.max_height = 4096;
> - fsize->stepwise.step_height = 2;
> + fsize->stepwise.min_height = RISP_MIN_HEIGHT;
> + fsize->stepwise.max_height = RISP_MAX_HEIGHT;
> + fsize->stepwise.step_height = 1u << RISP_SIZE_ALIGNMENT;
>
> return 0;
> }
> @@ -720,8 +727,9 @@ static const struct v4l2_pix_format_mplane risp_io_capture_default_format = {
> static void risp_io_capture_try_format(struct rcar_isp_core_io *io,
> struct v4l2_pix_format_mplane *pix)
> {
> - v4l_bound_align_image(&pix->width, 128, 5120, 2,
> - &pix->height, 128, 4096, 2, 0);
> + v4l_bound_align_image(&pix->width, RISP_MIN_WIDTH, RISP_MAX_WIDTH, RISP_SIZE_ALIGNMENT,
> + &pix->height, RISP_MIN_HEIGHT, RISP_MAX_HEIGHT, RISP_SIZE_ALIGNMENT,
> + 0);
>
> pix->field = V4L2_FIELD_NONE;
> pix->colorspace = V4L2_COLORSPACE_SRGB;
> @@ -824,13 +832,13 @@ static int risp_io_capture_enum_framesizes(struct file *file, void *fh,
>
> fsize->type = V4L2_FRMSIZE_TYPE_STEPWISE;
>
> - fsize->stepwise.min_width = 128;
> - fsize->stepwise.max_width = 5120;
> - fsize->stepwise.step_width = 2;
> + fsize->stepwise.min_width = RISP_MIN_WIDTH;
> + fsize->stepwise.max_width = RISP_MAX_WIDTH;
> + fsize->stepwise.step_width = 1u << RISP_SIZE_ALIGNMENT;
>
> - fsize->stepwise.min_height = 128;
> - fsize->stepwise.max_height = 4096;
> - fsize->stepwise.step_height = 2;
> + fsize->stepwise.min_height = RISP_MIN_HEIGHT;
> + fsize->stepwise.max_height = RISP_MAX_HEIGHT;
> + fsize->stepwise.step_height = 1u << RISP_SIZE_ALIGNMENT;
>
> return 0;
> }
> --
> 2.55.0
>
>
next prev parent reply other threads:[~2026-10-04 9:14 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-10-01 8:51 [PATCH v2] media: rcar-isp: ispcore: Fix inconsistent step sizes Barnabás Pőcze
2026-10-01 10:34 ` Sakari Ailus
2026-10-04 9:14 ` Jacopo Mondi [this message]
2026-10-05 8:55 ` Barnabás Pőcze
2026-10-05 9:24 ` Jacopo Mondi
2026-10-05 16:08 ` Niklas Söderlund
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=asIXMIVjbusYLbaR@zed \
--to=jacopo.mondi@ideasonboard.com \
--cc=barnabas.pocze+renesas@ideasonboard.com \
--cc=geert+renesas@glider.be \
--cc=jacopo.mondi+renesas@ideasonboard.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-media@vger.kernel.org \
--cc=linux-renesas-soc@vger.kernel.org \
--cc=magnus.damm@gmail.com \
--cc=mchehab@kernel.org \
--cc=niklas.soderlund@ragnatech.se \
--cc=sakari.ailus@linux.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 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.