Hi Dean,
Thanks for the patch.
On Fri, Dec 15, 2023 at 11:14:21AM -0800, Dean Anderson wrote:
> Adding V4L2 read-only control id for serial number as hardware
> does not support embedding the serial number in the USB device descriptors.
> Comment added noting v4l2_ctrl_handler_setup is not needed for this driver.
>
> Signed-off-by: Dean Anderson <dean@xxxxxxxxxxxx>
>
> ---
> drivers/media/usb/s2255/s2255drv.c | 52 +++++++++++++++++++++++++++++-
> 1 file changed, 51 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/media/usb/s2255/s2255drv.c b/drivers/media/usb/s2255/s2255drv.c
> index 3c2627712fe9..1f3961835711 100644
> --- a/drivers/media/usb/s2255/s2255drv.c
> +++ b/drivers/media/usb/s2255/s2255drv.c
> @@ -60,6 +60,7 @@
> #define S2255_MIN_BUFS 2
> #define S2255_SETMODE_TIMEOUT 500
> #define S2255_VIDSTATUS_TIMEOUT 350
> +#define S2255_MARKER_FIRMWARE cpu_to_le32(0xDDCCBBAAL)
It'd be nicer to convert the value obtained from the device to CPU
endianness rather than the other way around. But as this seems to be a
pattern already used in the driver, I'm fine with it here.
Given that the serial number itself is big endian, I wonder if the driver
is just implemented in a funny way and happens to work?
> #define S2255_MARKER_FRAME cpu_to_le32(0x2255DA4AL)
> #define S2255_MARKER_RESPONSE cpu_to_le32(0x2255ACACL)
> #define S2255_RESPONSE_SETMODE cpu_to_le32(0x01)
> @@ -323,6 +324,7 @@ struct s2255_buffer {
> #define S2255_V4L2_YC_ON 1
> #define S2255_V4L2_YC_OFF 0
> #define V4L2_CID_S2255_COLORFILTER (V4L2_CID_USER_S2255_BASE + 0)
> +#define V4L2_CID_S2255_SERIALNUM (V4L2_CID_USER_S2255_BASE + 1)
I'd document a new standardised control for this purpose. Any other
opinions?
>
> /* frame prefix size (sent once every frame) */
> #define PREFIX_SIZE 512
> @@ -1232,6 +1234,36 @@ static int s2255_s_ctrl(struct v4l2_ctrl *ctrl)
> return 0;
> }
>
> +/*
> + * serial number is not used in usb device descriptors.
> + * returns serial number from device, 0 if none found.
> + */
> +
> +static int s2255_g_serialnum(struct s2255_dev *dev)
> +{
> +#define S2255_SERIALNUM_NONE 0
No need for such a definition, just assign it to zero.
> +#define S2255_I2C_SIZE 16
> +#define S2255_I2C_SERIALNUM 0xa2
> +#define S2255_I2C_SERIALNUM_OFFSET 0x1ff0
> +#define S2255_VENDOR_READREG 0x22
> +
> + u8 *buf;
> + int serialnum = S2255_SERIALNUM_NONE;
u32?
> +
> + buf = kzalloc(S2255_I2C_SIZE, GFP_KERNEL);
Could this reside in the stack instead? It's just 16 bytes.
> + if (!buf)
> + return serialnum;
> +
> + s2255_vendor_req(dev, S2255_VENDOR_READREG, S2255_I2C_SERIALNUM_OFFSET,
> + S2255_I2C_SERIALNUM >> 1, buf, S2255_I2C_SIZE, 0);
It'd be nice to check for errors here.
> +
> + /* verify marker code */
> + if (*(__le32 *)buf == S2255_MARKER_FIRMWARE)
> + serialnum = (buf[12] << 24) + (buf[13] << 16) + (buf[14] << 8) + buf[15];
Maybe:
serialnum = be32_to_cpu(*(__be32 *)(buf + 12));
You could add a reference to the serial number as a function argument and
just return the error if one happens. This isn't exactly the same thing as
having serial number 0.
> + kfree(buf);
> + return serialnum;
> +}
> +
> static int vidioc_g_jpegcomp(struct file *file, void *priv,
> struct v4l2_jpegcompression *jc)
> {
> @@ -1581,6 +1613,17 @@ static const struct v4l2_ctrl_config color_filter_ctrl = {
> .def = 1,
> };
>
> +static const struct v4l2_ctrl_config v4l2_ctrl_serialnum = {
> + .ops = &s2255_ctrl_ops,
> + .name = "Serial Number",
> + .id = V4L2_CID_S2255_SERIALNUM,
> + .type = V4L2_CTRL_TYPE_INTEGER,
> + .max = 0x7fffffff,
> + .min = 0,
> + .step = 1,
> + .flags = V4L2_CTRL_FLAG_READ_ONLY,
> +};
> +
> static int s2255_probe_v4l(struct s2255_dev *dev)
> {
> int ret;
> @@ -1588,6 +1631,7 @@ static int s2255_probe_v4l(struct s2255_dev *dev)
> int cur_nr = video_nr;
> struct s2255_vc *vc;
> struct vb2_queue *q;
> + struct v4l2_ctrl_config tmp = v4l2_ctrl_serialnum;
>
> ret = v4l2_device_register(&dev->interface->dev, &dev->v4l2_dev);
> if (ret)
> @@ -1598,7 +1642,7 @@ static int s2255_probe_v4l(struct s2255_dev *dev)
> vc = &dev->vc[i];
> INIT_LIST_HEAD(&vc->buf_list);
>
> - v4l2_ctrl_handler_init(&vc->hdl, 6);
> + v4l2_ctrl_handler_init(&vc->hdl, 7);
> v4l2_ctrl_new_std(&vc->hdl, &s2255_ctrl_ops,
> V4L2_CID_BRIGHTNESS, -127, 127, 1, DEF_BRIGHT);
> v4l2_ctrl_new_std(&vc->hdl, &s2255_ctrl_ops,
> @@ -1615,6 +1659,8 @@ static int s2255_probe_v4l(struct s2255_dev *dev)
> (dev->pid != 0x2257 || vc->idx <= 1))
> v4l2_ctrl_new_custom(&vc->hdl, &color_filter_ctrl,
> NULL);
> + tmp.def = s2255_g_serialnum(dev);
> + v4l2_ctrl_new_custom(&vc->hdl, &tmp, NULL);
> if (vc->hdl.error) {
> ret = vc->hdl.error;
> v4l2_ctrl_handler_free(&vc->hdl);
> @@ -2306,6 +2352,10 @@ static int s2255_probe(struct usb_interface *interface,
> retval = s2255_board_init(dev);
> if (retval)
> goto errorBOARDINIT;
> + /*
> + * v4l2_ctrl_handler_setup is not required.
> + * V4L2 controls initialized when firmware is loaded.
> + */
I'd drop the comment.
> s2255_fwload_start(dev);
> /* loads v4l specific */
> retval = s2255_probe_v4l(dev);
--
Regards,
Sakari Ailus
