On 19/04/2023 11:04, Sakari Ailus wrote:
Hi Bingbu,
On Wed, Apr 19, 2023 at 03:44:25PM +0800, Bingbu Cao wrote:
Sakari,
On 4/12/23 8:19 PM, Sakari Ailus wrote:
Hi Bingbu,
On Wed, Apr 12, 2023 at 07:40:19PM +0800, Bingbu Cao wrote:
+static int ov01a10_set_format(struct v4l2_subdev *sd,
+ struct v4l2_subdev_state *sd_state,
+ struct v4l2_subdev_format *fmt)
+{
+ struct ov01a10 *ov01a10 = to_ov01a10(sd);
+ const struct ov01a10_mode *mode;
+ s32 vblank_def, h_blank;
+
+ mode = v4l2_find_nearest_size(supported_modes,
+ ARRAY_SIZE(supported_modes), width,
+ height, fmt->format.width,
+ fmt->format.height);
+
+ mutex_lock(&ov01a10->mutex);
+ ov01a10_update_pad_format(mode, &fmt->format);
Could you switch to the sub-device state? That is now the preferred way to
serialise access to e.g. the format.
See e.g.
<URL:https://git.kernel.org/pub/scm/linux/kernel/git/tomba/linux.git/tree/drivers/media/i2c/ov1063x.c?h=streams/work-v16>.
The control handler's mutex doubles as a sub-device state mutex.
Is it fine to use v4l2_subdev_get_fmt()? Or will it be deprecated soon?
Sure, it's fine to use it.
I tried to use the v4l2_subdev_state_get_stream_format() in set_fmt, but I
see that the pad format of camera sensor was missing in media information.
- entity 105: ov01a10 3-0036 (1 pad, 1 link)
type V4L2 subdev subtype Sensor flags 0
device node name /dev/v4l-subdev8
pad0: Source
-> "Intel IPU6 CSI2 2":0 [ENABLED]
And the link validation cannot work as expected, is there something
missing? My kernel head is Linux 6.3-rc2.
static int ov01a10_set_format(struct v4l2_subdev *sd,
struct v4l2_subdev_state *sd_state,
struct v4l2_subdev_format *fmt)
{
...
format = v4l2_subdev_state_get_stream_format(sd_state, fmt->pad,
fmt->stream);
if (!format) {
dev_err(&client->dev, "Failed to get stream format\n");
return -EINVAL;
}
...
ov01a10_update_pad_format(mode, &fmt->format);
*format = fmt->format;
return 0;
}
v4l2_subdev_state_get_stream_format() will access streams and I presume you
won't have any in this case (not even streams support right now as it's a
sensor driver with a single sub-device).
This perhaps gives a better example on how to migrate to subdev state:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/drivers/media/i2c/imx290.c?id=a2514b9a634ac0a2cfbc329822b8fb58ffe23a80
Tomi
