linux-kernel - Re: [PATCH v8 7/8] media: i2c: ov9282: implement try_ctrl for strobe

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <vcrbxvgco2vuhkdzlbcbirl67otjav26wv4gf6btjd7vkzdyhd@qcrpbdd54p2z>
Date: Wed, 12 Nov 2025 09:01:34 +0100
From: Richard Leitner <richard.leitner@...ux.dev>
To: Sakari Ailus <sakari.ailus@...ux.intel.com>
Cc: Dave Stevenson <dave.stevenson@...pberrypi.com>, 
	Mauro Carvalho Chehab <mchehab@...nel.org>, Lee Jones <lee@...nel.org>, Pavel Machek <pavel@...nel.org>, 
	Laurent Pinchart <laurent.pinchart@...asonboard.com>, linux-media@...r.kernel.org, linux-kernel@...r.kernel.org, 
	linux-leds@...r.kernel.org, Hans Verkuil <hverkuil@...nel.org>
Subject: Re: [PATCH v8 7/8] media: i2c: ov9282: implement try_ctrl for
 strobe_duration

Hi Sakari,

thanks for the quick feedback!

On Mon, Nov 10, 2025 at 09:55:18AM +0200, Sakari Ailus wrote:
> Hi Richard,
> 
> On Tue, Nov 04, 2025 at 03:30:58PM +0100, Richard Leitner wrote:
> > As the granularity of the hardware supported values is lower than the
> > control value, implement a try_ctrl() function for
> > V4L2_CID_FLASH_DURATION. This function calculates the nearest possible
> > µs strobe duration for the given value and returns it back to the
> > caller.
> > 
> > Signed-off-by: Richard Leitner <richard.leitner@...ux.dev>
> > ---
> >  drivers/media/i2c/ov9282.c | 55 +++++++++++++++++++++++++++++++++++++++++++---
> >  1 file changed, 52 insertions(+), 3 deletions(-)
> > 
> > diff --git a/drivers/media/i2c/ov9282.c b/drivers/media/i2c/ov9282.c
> > index 26296dc1d1b95..7aa74feb3ee44 100644
> > --- a/drivers/media/i2c/ov9282.c
> > +++ b/drivers/media/i2c/ov9282.c
> > @@ -128,6 +128,8 @@
> >  #define OV9282_REG_MIN		0x00
> >  #define OV9282_REG_MAX		0xfffff
> >  
> > +#define OV9282_STROBE_SPAN_FACTOR	192
> > +
> >  static const char * const ov9282_supply_names[] = {
> >  	"avdd",		/* Analog power */
> >  	"dovdd",	/* Digital I/O power */
> > @@ -691,9 +693,8 @@ static int ov9282_set_ctrl_flash_strobe_oe(struct ov9282 *ov9282, bool enable)
> >  	return ov9282_write_reg(ov9282, OV9282_REG_OUTPUT_ENABLE6, 1, current_val);
> >  }
> >  
> > -static int ov9282_set_ctrl_flash_duration(struct ov9282 *ov9282, u32 value)
> > +static u32 ov9282_us_to_flash_duration(struct ov9282 *ov9282, u32 value)
> >  {
> > -	int ret;
> >  	/*
> >  	 * Calculate "strobe_frame_span" increments from a given value (µs).
> >  	 * This is quite tricky as "The step width of shift and span is
> > @@ -703,7 +704,28 @@ static int ov9282_set_ctrl_flash_duration(struct ov9282 *ov9282, u32 value)
> >  	 * The formula below is interpolated from different modes/framerates
> >  	 * and should work quite well for most settings.
> >  	 */
> > -	u32 val = value * 192 / (ov9282->cur_mode->width + ov9282->hblank_ctrl->val);
> > +	u32 frame_width = ov9282->cur_mode->width + ov9282->hblank_ctrl->val;
> > +
> > +	return value * OV9282_STROBE_SPAN_FACTOR / frame_width;
> > +}
> > +
> > +static u32 ov9282_flash_duration_to_us(struct ov9282 *ov9282, u32 value)
> > +{
> > +	/*
> > +	 * As the calculation in ov9282_us_to_flash_duration uses an integer
> > +	 * divison calculate in ns here to get more precision. Then check if
> > +	 * we need to compensate that divison by incrementing the µs result.
> > +	 */
> > +	u32 frame_width = ov9282->cur_mode->width + ov9282->hblank_ctrl->val;
> > +	u64 ns = value * 1000 * frame_width / OV9282_STROBE_SPAN_FACTOR;
> > +
> > +	return DIV_ROUND_UP(ns, 1000);
> 
> Can ns even exceed U32_MAX? It doesn't seem to be useful to multiply and
> then divide by 1000.

You're right. That multiply/divide by 1000 is obsolete. I will get rid
of it by calculating directly using DIV_ROUND_UP().

This will make the largest value "value * frame_width", which shouldn't
exceed U32_MAX in any real-world use-case IMHO...

> 
> I guess this warrants v9, please fix the other issue for that, too.
> 

Sure. Will send a v9 with both fixes later today. Thanks!

> > +}
> > +
> > +static int ov9282_set_ctrl_flash_duration(struct ov9282 *ov9282, u32 value)
> > +{
> > +	int ret;
> > +	u32 val = ov9282_us_to_flash_duration(ov9282, value);
> >  
> >  	ret = ov9282_write_reg(ov9282, OV9282_REG_STROBE_FRAME_SPAN, 1,
> >  			       (val >> 24) & 0xff);
> > @@ -798,9 +820,36 @@ static int ov9282_set_ctrl(struct v4l2_ctrl *ctrl)
> >  	return ret;
> >  }
> >  
> > +static int ov9282_try_ctrl(struct v4l2_ctrl *ctrl)
> > +{
> > +	struct ov9282 *ov9282 =
> > +		container_of_const(ctrl->handler, struct ov9282, ctrl_handler);
> > +
> > +	if (ctrl->id == V4L2_CID_FLASH_DURATION) {
> > +		u32 us = ctrl->val;
> > +		u32 fd = ov9282_us_to_flash_duration(ov9282, us);
> > +
> > +		/* get nearest strobe_duration value */
> > +		u32 us0 = ov9282_flash_duration_to_us(ov9282, fd);
> > +		u32 us1 = ov9282_flash_duration_to_us(ov9282, fd + 1);
> > +
> > +		if (abs(us1 - us) < abs(us - us0))
> > +			ctrl->val = us1;
> > +		else
> > +			ctrl->val = us0;
> > +
> > +		if (us != ctrl->val)
> > +			dev_dbg(ov9282->dev, "using next valid strobe_duration %u instead of %u\n",
> > +				ctrl->val, us);
> > +	}
> > +
> > +	return 0;
> > +}
> > +
> >  /* V4l2 subdevice control ops*/
> >  static const struct v4l2_ctrl_ops ov9282_ctrl_ops = {
> >  	.s_ctrl = ov9282_set_ctrl,
> > +	.try_ctrl = ov9282_try_ctrl,
> >  };
> >  
> >  /**
> > 
> 
> -- 
> Regards,
> 
> Sakari Ailus

regards;rl