On 2023-12-12 at 00:08 +02, Laurent Pinchart <laurent.pinchart@xxxxxxxxxxxxxxxx> wrote:
> Hi Mikhail,
>
> Thank you for the patch.
>
> On Mon, Dec 11, 2023 at 08:50:13PM +0300, Mikhail Rudenko wrote:
>> Make horizontal blanking configurable. To do so, set HTS register
>> according to the requested horizontal blanking in ov4689_set_ctrl
>> instead of the register table. Default HTS value is not changed by
>> this patch. Minimal HTS value is found experimentally and corresponds
>> to 90 fps framerate at minimum vertical blanking. Real HTS value is
>> the register value multiplied by 4.
>>
>> Signed-off-by: Mikhail Rudenko <mike.rudenko@xxxxxxxxx>
>> ---
>> drivers/media/i2c/ov4689.c | 33 +++++++++++++++++++++------------
>> 1 file changed, 21 insertions(+), 12 deletions(-)
>>
>> diff --git a/drivers/media/i2c/ov4689.c b/drivers/media/i2c/ov4689.c
>> index 9fa06941a0e5..67d4004bdcfb 100644
>> --- a/drivers/media/i2c/ov4689.c
>> +++ b/drivers/media/i2c/ov4689.c
>> @@ -30,7 +30,6 @@
>> #define OV4689_REG_EXPOSURE CCI_REG24(0x3500)
>> #define OV4689_EXPOSURE_MIN 4
>> #define OV4689_EXPOSURE_STEP 1
>> -#define OV4689_VTS_MAX 0x7fff
>>
>> #define OV4689_REG_GAIN CCI_REG16(0x3508)
>> #define OV4689_GAIN_STEP 1
>> @@ -41,6 +40,11 @@
>> #define OV4689_TEST_PATTERN_DISABLE 0x0
>>
>> #define OV4689_REG_VTS CCI_REG16(0x380e)
>> +#define OV4689_VTS_MAX 0x7fff
>> +
>> +#define OV4689_REG_HTS CCI_REG16(0x380c)
>> +#define OV4689_HTS_DIVIDER 4
>> +#define OV4689_HTS_MAX 0x7fff
>
> Could you move this just before REG_VTS to keep registers sorted by
> address ?
Ack, will fix in v2.
>>
>> #define OV4689_LANES 4
>> #define OV4689_XVCLK_FREQ 24000000
>> @@ -61,6 +65,7 @@ struct ov4689_mode {
>> u32 width;
>> u32 height;
>> u32 hts_def;
>> + u32 hts_min;
>> u32 vts_def;
>> u32 exp_def;
>> u32 pixel_rate;
>> @@ -104,7 +109,7 @@ struct ov4689_gain_range {
>>
>> /*
>> * Xclk 24Mhz
>> - * max_framerate 30fps
>> + * max_framerate 90fps
>> * mipi_datarate per lane 1008Mbps
>> */
>> static const struct cci_reg_sequence ov4689_2688x1520_regs[] = {
>> @@ -175,8 +180,6 @@ static const struct cci_reg_sequence ov4689_2688x1520_regs[] = {
>> /* Timing control */
>> {CCI_REG8(0x3801), 0x08}, /* H_CROP_START_L h_crop_start[7:0] = 0x08 */
>> {CCI_REG8(0x3805), 0x97}, /* H_CROP_END_L h_crop_end[7:0] = 0x97 */
>> - {CCI_REG8(0x380c), 0x0a}, /* TIMING_HTS_H hts[14:8] = 0x0a */
>> - {CCI_REG8(0x380d), 0x0e}, /* TIMING_HTS_L hts[7:0] = 0x0e */
>> {CCI_REG8(0x3811), 0x08}, /* H_WIN_OFF_L h_win_off[7:0] = 0x08*/
>> {CCI_REG8(0x3813), 0x04}, /* V_WIN_OFF_L v_win_off[7:0] = 0x04 */
>> {CCI_REG8(0x3819), 0x01}, /* VSYNC_END_L vsync_end_point[7:0] = 0x01 */
>> @@ -237,7 +240,8 @@ static const struct ov4689_mode supported_modes[] = {
>> .crop_top = 8,
>> .crop_left = 16,
>> .exp_def = 1536,
>> - .hts_def = 4 * 2574,
>> + .hts_def = 10296,
>> + .hts_min = 3432,
>> .vts_def = 1554,
>> .pixel_rate = 480000000,
>> .reg_list = ov4689_2688x1520_regs,
>> @@ -596,6 +600,11 @@ static int ov4689_set_ctrl(struct v4l2_ctrl *ctrl)
>> case V4L2_CID_TEST_PATTERN:
>> ret = ov4689_enable_test_pattern(ov4689, val);
>> break;
>> + case V4L2_CID_HBLANK:
>> + cci_write(regmap, OV4689_REG_HTS,
>> + (val + ov4689->cur_mode->width) /
>> + OV4689_HTS_DIVIDER, &ret);
>> + break;
>> default:
>> dev_warn(dev, "%s Unhandled id:0x%x, val:0x%x\n",
>> __func__, ctrl->id, val);
>> @@ -618,13 +627,13 @@ static int ov4689_initialize_controls(struct ov4689 *ov4689)
>> struct v4l2_ctrl_handler *handler;
>> const struct ov4689_mode *mode;
>> s64 exposure_max, vblank_def;
>> + s64 hblank_def, hblank_min;
>> struct v4l2_ctrl *ctrl;
>> - s64 h_blank_def;
>> int ret;
>>
>> handler = &ov4689->ctrl_handler;
>> mode = ov4689->cur_mode;
>> - ret = v4l2_ctrl_handler_init(handler, 10);
>> + ret = v4l2_ctrl_handler_init(handler, 11);
>
> The HBLANK control already exists, you're only changing how it is
> initialized. I see 8 controls being created by the driver directly,
> plus 2 created by v4l2_ctrl_new_fwnode_properties(), so I think 10 is a
> correct value here.
Yes, will adjust here and in the further patches.
> With these small issues addressed,
>
> Reviewed-by: Laurent Pinchart <laurent.pinchart@xxxxxxxxxxxxxxxx>
>
>> if (ret)
>> return ret;
>>
>> @@ -636,11 +645,11 @@ static int ov4689_initialize_controls(struct ov4689 *ov4689)
>> v4l2_ctrl_new_std(handler, NULL, V4L2_CID_PIXEL_RATE, 0,
>> mode->pixel_rate, 1, mode->pixel_rate);
>>
>> - h_blank_def = mode->hts_def - mode->width;
>> - ctrl = v4l2_ctrl_new_std(handler, NULL, V4L2_CID_HBLANK, h_blank_def,
>> - h_blank_def, 1, h_blank_def);
>> - if (ctrl)
>> - ctrl->flags |= V4L2_CTRL_FLAG_READ_ONLY;
>> + hblank_def = mode->hts_def - mode->width;
>> + hblank_min = mode->hts_min - mode->width;
>> + v4l2_ctrl_new_std(handler, &ov4689_ctrl_ops, V4L2_CID_HBLANK,
>> + hblank_min, OV4689_HTS_MAX - mode->width,
>> + OV4689_HTS_DIVIDER, hblank_def);
>>
>> vblank_def = mode->vts_def - mode->height;
>> v4l2_ctrl_new_std(handler, &ov4689_ctrl_ops, V4L2_CID_VBLANK,
--
Best regards,
Mikhail Rudenko
